{"id":24849,"date":"2026-06-30T16:32:26","date_gmt":"2026-06-30T09:32:26","guid":{"rendered":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/?p=24849"},"modified":"2026-07-16T13:37:31","modified_gmt":"2026-07-16T06:37:31","slug":"prompt-engineer-la-gi","status":"publish","type":"post","link":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/tu-van-nghe-nghiep\/prompt-engineer-la-gi","title":{"rendered":"Prompt Engineer L\u00e0 G\u00ec? K\u1ef9 N\u0103ng, L\u01b0\u01a1ng v\u00e0 L\u1ed9 Tr\u00ecnh V\u00e0o Ngh\u1ec1 AI"},"content":{"rendered":"

Trong b\u1ed1i c\u1ea3nh AI b\u00f9ng n\u1ed5 to\u00e0n c\u1ea7u, th\u1ecb tr\u01b0\u1eddng lao \u0111\u1ed9ng xu\u1ea5t hi\u1ec7n v\u1ecb tr\u00ed ch\u01b0a t\u1eebng c\u00f3 trong b\u1ea3ng m\u00f4 t\u1ea3 ngh\u1ec1 nghi\u1ec7p truy\u1ec1n th\u1ed1ng. Prompt Engineer<\/strong> \u2014 k\u1ef9 s\u01b0 thi\u1ebft k\u1ebf c\u00e2u l\u1ec7nh AI \u2014 tr\u1edf th\u00e0nh vai tr\u00f2 \u0111\u01b0\u1ee3c s\u0103n \u0111\u00f3n nh\u1ea5t t\u1ea1i c\u00f4ng ty c\u00f4ng ngh\u1ec7, ng\u00e2n h\u00e0ng v\u00e0 startup Vi\u1ec7t Nam l\u1eabn qu\u1ed1c t\u1ebf, nh\u1edd l\u00e0n s\u00f3ng tri\u1ec3n khai m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef l\u1edbn.<\/p>\n

\"Prompt<\/p>\n

\nT\u1ed5ng quan nhanh v\u1ec1 Prompt Engineer<\/strong>\n

\u2013 World Economic Forum “Future of Jobs 2023” x\u1ebfp AI Prompt Engineering v\u00e0o top 10 k\u1ef9 n\u0103ng m\u1edbi nh\u1ea5t c\u1ea7n c\u00f3 \u0111\u1ebfn n\u0103m 2027.<\/p>\n

\u2013 M\u1ee9c l\u01b0\u01a1ng to\u00e0n c\u1ea7u: $75,000\u2013$300,000\/n\u0103m; Vi\u1ec7t Nam: $1,200\u2013$4,000\/th\u00e1ng t\u00f9y c\u1ea5p \u0111\u1ed9 kinh nghi\u1ec7m.<\/p>\n

\u2013 C\u00e1c c\u00f4ng ty Vi\u1ec7t Nam \u0111ang tuy\u1ec3n: VinAI Research, Techcombank AI Lab, FPT AI Center, KMS Technology, Base.vn.<\/p>\n

\u2013 Quy\u1ebft \u0111\u1ecbnh 127\/Q\u0110-TTg v\u1ec1 Chi\u1ebfn l\u01b0\u1ee3c AI Qu\u1ed1c gia m\u1edf ra nhu c\u1ea7u nh\u00e2n s\u1ef1 AI t\u1ea1i c\u1ea3 khu v\u1ef1c c\u00f4ng l\u1eabn t\u01b0 nh\u00e2n.<\/p>\n<\/div>\n

1. Prompt Engineer l\u00e0 g\u00ec?<\/h2>\n

Prompt Engineer (K\u1ef9 s\u01b0 Prompt) l\u00e0 chuy\u00ean gia thi\u1ebft k\u1ebf, tinh ch\u1ec9nh v\u00e0 t\u1ed1i \u01b0u h\u00f3a c\u00e2u l\u1ec7nh \u0111\u1ea7u v\u00e0o (prompt) nh\u1eb1m khai th\u00e1c t\u1ed1i \u0111a kh\u1ea3 n\u0103ng c\u1ee7a c\u00e1c m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef l\u1edbn \u2014 Large Language Model (LLM) \u2014 nh\u01b0 GPT-4, Claude hay Gemini. \u0110\u00e2y kh\u00f4ng \u0111\u01a1n gi\u1ea3n l\u00e0 “h\u1ecfi ChatGPT cho \u0111\u00fang c\u00e1ch” m\u00e0 l\u00e0 k\u1ef9 thu\u1eadt nghi\u00eam t\u00fac \u0111\u00f2i h\u1ecfi t\u01b0 duy h\u1ec7 th\u1ed1ng v\u00e0 th\u1eed nghi\u1ec7m c\u00f3 ph\u01b0\u01a1ng ph\u00e1p, gi\u00fap doanh nghi\u1ec7p bi\u1ebfn AI th\u00e0nh c\u00f4ng c\u1ee5 s\u1ea3n xu\u1ea5t n\u1ed9i dung, ph\u00e2n t\u00edch v\u0103n b\u1ea3n ph\u00e1p l\u00fd ho\u1eb7c h\u1ed7 tr\u1ee3 kh\u00e1ch h\u00e0ng v\u1edbi \u0111\u1ed9 ch\u00ednh x\u00e1c cao.<\/p>\n

\n

“Vi\u1ebft prompt t\u1ed1t kh\u00f4ng kh\u00e1c g\u00ec l\u1eadp tr\u00ecnh \u2014 b\u1ea1n ph\u1ea3i hi\u1ec3u m\u00e1y m\u00f3c \u0111ang ‘ngh\u0129’ nh\u01b0 th\u1ebf n\u00e0o v\u00e0 ph\u1ea3i ki\u1ec3m so\u00e1t qu\u00e1 tr\u00ecnh \u0111\u00f3 b\u1eb1ng ng\u00f4n ng\u1eef t\u1ef1 nhi\u00ean.” \u2014 Andrej Karpathy, c\u1ef1u Gi\u00e1m \u0111\u1ed1c AI t\u1ea1i Tesla v\u00e0 OpenAI, 2023.<\/p>\n<\/blockquote>\n

1. Ph\u00e2n bi\u1ec7t Prompt Engineer, Prompt Engineering v\u00e0 AI Engineer<\/h3>\n

Ba kh\u00e1i ni\u1ec7m th\u01b0\u1eddng b\u1ecb nh\u1ea7m l\u1eabn v\u1edbi nhau nh\u01b0ng c\u00f3 ranh gi\u1edbi kh\u00e1 r\u00f5 r\u00e0ng v\u1ec1 ph\u1ea1m vi v\u00e0 chi\u1ec1u s\u00e2u k\u1ef9 thu\u1eadt:<\/p>\n

\u2013 Prompt Engineering<\/strong> l\u00e0 ngh\u1ec7 thu\u1eadt v\u00e0 khoa h\u1ecdc thi\u1ebft k\u1ebf \u0111\u1ea7u v\u00e0o (input) cho m\u00f4 h\u00ecnh AI \u0111\u1ec3 t\u1ea1o ra \u0111\u1ea7u ra (output) ph\u00f9 h\u1ee3p m\u1ee5c ti\u00eau kinh doanh \u2014 \u0111\u00e2y l\u00e0 k\u1ef9 n\u0103ng, kh\u00f4ng ph\u1ea3i ch\u1ee9c danh ngh\u1ec1 nghi\u1ec7p.<\/p>\n

\u2013 Prompt Engineer<\/strong> l\u00e0 ng\u01b0\u1eddi h\u00e0nh ngh\u1ec1 prompt engineering \u1edf c\u1ea5p \u0111\u1ed9 chuy\u00ean nghi\u1ec7p, th\u01b0\u1eddng k\u1ebft h\u1ee3p Python v\u00e0 c\u00e1c framework nh\u01b0 LangChain \u0111\u1ec3 t\u1ef1 \u0111\u1ed9ng h\u00f3a quy tr\u00ecnh.<\/p>\n

\u2013 AI Engineer<\/strong> l\u00e0 k\u1ef9 s\u01b0 x\u00e2y d\u1ef1ng h\u1ec7 th\u1ed1ng AI \u0111\u1ea7u cu\u1ed1i, bao g\u1ed3m hu\u1ea5n luy\u1ec7n m\u00f4 h\u00ecnh, tri\u1ec3n khai s\u1ea3n xu\u1ea5t v\u00e0 qu\u1ea3n l\u00fd MLOps \u2014 vai tr\u00f2 k\u1ef9 thu\u1eadt n\u1eb7ng v\u00e0 \u0111\u00f2i h\u1ecfi n\u1ec1n t\u1ea3ng to\u00e1n h\u1ecdc s\u00e2u h\u01a1n.<\/p>\n

2. Prompt Engineer l\u00e0m g\u00ec trong c\u00f4ng vi\u1ec7c h\u00e0ng ng\u00e0y?<\/h2>\n

C\u00f4ng vi\u1ec7c h\u00e0ng ng\u00e0y c\u1ee7a Prompt Engineer xoay quanh m\u1ed9t tr\u1ee5c c\u1ed1t l\u00f5i: l\u00e0m cho m\u00f4 h\u00ecnh AI t\u1ea1o ra k\u1ebft qu\u1ea3 s\u1eed d\u1ee5ng \u0111\u01b0\u1ee3c trong th\u1ef1c t\u1ebf. \u1ede startup, ng\u01b0\u1eddi n\u00e0y thi\u1ebft k\u1ebf chatbot v\u00e0 \u0111\u00e1nh gi\u00e1 ch\u1ea5t l\u01b0\u1ee3ng \u0111\u1ea7u ra; \u1edf ng\u00e2n h\u00e0ng, c\u00f4ng vi\u1ec7c t\u1eadp trung v\u00e0o t\u00edch h\u1ee3p API v\u00e0 ki\u1ec3m th\u1eed h\u1ec7 th\u1ed1ng \u2014 \u0111\u1ec1u \u0111\u00f2i h\u1ecfi t\u01b0 duy v\u1eeba s\u00e1ng t\u1ea1o v\u1eeba ph\u00e2n t\u00edch, bi\u1ebft c\u00e1ch “n\u00f3i chuy\u1ec7n” v\u1edbi AI v\u00e0 \u0111o l\u01b0\u1eddng k\u1ebft qu\u1ea3 b\u1eb1ng s\u1ed1 li\u1ec7u c\u1ee5 th\u1ec3.<\/p>\n

1. C\u00e1c nhi\u1ec7m v\u1ee5 k\u1ef9 thu\u1eadt c\u1ed1t l\u00f5i<\/h3>\n

\u2013 Thi\u1ebft k\u1ebf v\u00e0 ki\u1ec3m th\u1eed prompt theo k\u1ef9 thu\u1eadt few-shot, zero-shot v\u00e0 Chain-of-Thought (CoT) \u0111\u1ec3 c\u1ea3i thi\u1ec7n \u0111\u1ed9 ch\u00ednh x\u00e1c \u0111\u1ea7u ra theo t\u1eebng t\u00e1c v\u1ee5 c\u1ee5 th\u1ec3.<\/p>\n

\u2013 T\u00edch h\u1ee3p m\u00f4 h\u00ecnh AI v\u00e0o \u1ee9ng d\u1ee5ng qua API c\u1ee7a OpenAI, Anthropic ho\u1eb7c Google Vertex AI b\u1eb1ng Python k\u1ebft h\u1ee3p LangChain ho\u1eb7c LlamaIndex.<\/p>\n

\u2013 X\u00e2y d\u1ef1ng pipeline RAG (Retrieval-Augmented Generation) \u0111\u1ec3 m\u00f4 h\u00ecnh AI tr\u1ea3 l\u1eddi d\u1ef1a tr\u00ean d\u1eef li\u1ec7u n\u1ed9i b\u1ed9 doanh nghi\u1ec7p thay v\u00ec ch\u1ec9 d\u1ef1a v\u00e0o ki\u1ebfn th\u1ee9c hu\u1ea5n luy\u1ec7n s\u1eb5n.<\/p>\n

\u2013 \u0110\u00e1nh gi\u00e1 ch\u1ea5t l\u01b0\u1ee3ng \u0111\u1ea7u ra theo c\u00e1c ch\u1ec9 s\u1ed1: \u0111\u1ed9 ch\u00ednh x\u00e1c, t\u00ednh nh\u1ea5t qu\u00e1n, an to\u00e0n n\u1ed9i dung (content safety) v\u00e0 ph\u00e1t hi\u1ec7n hallucination.<\/p>\n

\u2013 Ph\u1ed1i h\u1ee3p v\u1edbi Product Manager, Business Analyst \u0111\u1ec3 d\u1ecbch y\u00eau c\u1ea7u kinh doanh th\u00e0nh \u0111\u1eb7c t\u1ea3 k\u1ef9 thu\u1eadt cho m\u00f4 h\u00ecnh AI ho\u1ea1t \u0111\u1ed9ng \u0111\u00fang ng\u1eef c\u1ea3nh.<\/p>\n

\u2013 T\u00e0i li\u1ec7u h\u00f3a th\u01b0 vi\u1ec7n prompt v\u00e0 x\u00e2y d\u1ef1ng quy tr\u00ecnh ki\u1ec3m th\u1eed t\u1ef1 \u0111\u1ed9ng (prompt regression testing) khi h\u1ec7 th\u1ed1ng \u0111\u01b0\u1ee3c c\u1eadp nh\u1eadt.<\/p>\n

Nhu c\u1ea7u tuy\u1ec3n d\u1ee5ng nh\u00e2n s\u1ef1 t\u00edch h\u1ee3p AI v\u00e0o s\u1ea3n ph\u1ea9m ph\u1ea7n m\u1ec1m \u0111ang t\u0103ng nhanh tr\u00ean kh\u1eafp Vi\u1ec7t Nam \u2014 xem c\u00e1c v\u1ecb tr\u00ed m\u1edbi nh\u1ea5t t\u1ea1i chuy\u00ean m\u1ee5c vi\u1ec7c l\u00e0m CNTT ph\u1ea7n m\u1ec1m<\/a> \u0111\u1ec3 n\u1eafm b\u1eaft y\u00eau c\u1ea7u tuy\u1ec3n d\u1ee5ng th\u1ef1c t\u1ebf t\u1eeb nh\u00e0 tuy\u1ec3n d\u1ee5ng trong ng\u00e0nh.<\/p>\n

\"Prompt<\/p>\n

3. K\u1ef9 n\u0103ng c\u1ea7n c\u00f3 \u0111\u1ec3 tr\u1edf th\u00e0nh Prompt Engineer<\/h2>\n

Prompt Engineer l\u00e0 v\u1ecb tr\u00ed giao thoa gi\u1eefa k\u1ef9 thu\u1eadt v\u00e0 ng\u00f4n ng\u1eef. Ng\u01b0\u1eddi t\u1eeb ng\u00e0nh ng\u00f4n ng\u1eef h\u1ecdc hay vi\u1ebft n\u1ed9i dung c\u00f3 l\u1ee3i th\u1ebf t\u01b0 duy di\u1ec5n \u0111\u1ea1t; l\u1eadp tr\u00ecnh vi\u00ean c\u00f3 t\u01b0 duy h\u1ec7 th\u1ed1ng \u0111\u1ec3 h\u1ecdc ph\u1ea7n k\u1ef9 thu\u1eadt nhanh h\u01a1n \u2014 v\u00e0 th\u1ecb tr\u01b0\u1eddng \u0111ang c\u1ea7n c\u1ea3 hai nh\u00f3m, \u01b0u ti\u00ean ng\u01b0\u1eddi t\u00ecm \u0111\u01b0\u1ee3c \u0111i\u1ec3m giao nhau c\u1ee7a ch\u00fang.<\/p>\n

1. K\u1ef9 n\u0103ng k\u1ef9 thu\u1eadt (Hard Skills)<\/h3>\n

\u2013 Python<\/strong> l\u00e0 k\u1ef9 n\u0103ng l\u1eadp tr\u00ecnh n\u1ec1n t\u1ea3ng, d\u00f9ng \u0111\u1ec3 vi\u1ebft script t\u1ef1 \u0111\u1ed9ng h\u00f3a prompt, x\u1eed l\u00fd d\u1eef li\u1ec7u JSON v\u00e0 g\u1ecdi API \u2014 kh\u00f4ng c\u1ea7n \u1edf c\u1ea5p backend chuy\u00ean s\u00e2u nh\u01b0ng ph\u1ea3i th\u00e0nh th\u1ea1o th\u01b0 vi\u1ec7n `openai`, `anthropic` v\u00e0 `requests`.<\/p>\n

\u2013 K\u1ef9 thu\u1eadt prompting n\u00e2ng cao<\/strong> bao g\u1ed3m Chain-of-Thought (CoT), Tree-of-Thought, few-shot learning, zero-shot prompting, role prompting v\u00e0 structured output v\u1edbi JSON schema.<\/p>\n

\u2013 RAG v\u00e0 Vector Database<\/strong> l\u00e0 k\u1ef9 n\u0103ng x\u00e2y d\u1ef1ng h\u1ec7 th\u1ed1ng t\u00ecm ki\u1ebfm ng\u1eef ngh\u0129a v\u1edbi FAISS, Pinecone ho\u1eb7c Chroma \u0111\u1ec3 l\u00e0m gi\u00e0u ng\u1eef c\u1ea3nh cho LLM khi truy v\u1ea5n d\u1eef li\u1ec7u n\u1ed9i b\u1ed9.<\/p>\n

\u2013 LangChain v\u00e0 LlamaIndex<\/strong> l\u00e0 hai framework ph\u1ed5 bi\u1ebfn nh\u1ea5t hi\u1ec7n nay \u0111\u1ec3 x\u00e2y d\u1ef1ng \u1ee9ng d\u1ee5ng LLM c\u00f3 chu\u1ed7i logic ph\u1ee9c t\u1ea1p v\u00e0 kh\u1ea3 n\u0103ng k\u1ebft n\u1ed1i \u0111a ngu\u1ed3n d\u1eef li\u1ec7u.<\/p>\n

\u2013 Hi\u1ec3u c\u01a1 b\u1ea3n v\u1ec1 LLM<\/strong> bao g\u1ed3m tokenization, temperature, top-p sampling, context window v\u00e0 c\u00e1ch ki\u1ec3m so\u00e1t hallucination b\u1eb1ng grounding v\u00e0 system prompt.<\/p>\n

2. K\u1ef9 n\u0103ng m\u1ec1m v\u00e0 ng\u00f4n ng\u1eef<\/h3>\n

\u2013 Ti\u1ebfng Anh t\u1eeb B2 tr\u1edf l\u00ean l\u00e0 c\u1ea7n thi\u1ebft \u0111\u1ec3 \u0111\u1ecdc t\u00e0i li\u1ec7u k\u1ef9 thu\u1eadt, vi\u1ebft prompt hi\u1ec7u qu\u1ea3 v\u00e0 theo d\u00f5i c\u1eadp nh\u1eadt t\u1eeb c\u1ed9ng \u0111\u1ed3ng AI qu\u1ed1c t\u1ebf v\u1ed1n ch\u1ee7 y\u1ebfu d\u00f9ng ti\u1ebfng Anh.<\/p>\n

\u2013 T\u01b0 duy ph\u1ea3n bi\u1ec7n v\u00e0 ph\u00e2n t\u00edch l\u1ed7i l\u00e0 n\u0103ng l\u1ef1c nh\u1eadn bi\u1ebft v\u00ec sao m\u00f4 h\u00ecnh cho k\u1ebft qu\u1ea3 sai v\u00e0 \u0111i\u1ec1u ch\u1ec9nh prompt c\u00f3 h\u1ec7 th\u1ed1ng thay v\u00ec th\u1eed ng\u1eabu nhi\u00ean.<\/p>\n

\u2013 K\u1ef9 n\u0103ng vi\u1ebft r\u00f5 r\u00e0ng, c\u00f3 c\u1ea5u tr\u00fac l\u00e0 n\u1ec1n t\u1ea3ng tr\u1ef1c ti\u1ebfp \u0111\u1ec3 thi\u1ebft k\u1ebf system prompt v\u00e0 instruction ch\u1ea5t l\u01b0\u1ee3ng cao cho m\u00f4 h\u00ecnh AI.<\/p>\n

N\u1ebfu b\u1ea1n \u0111ang c\u00e2n nh\u1eafc gi\u1eefa c\u00e1c vai tr\u00f2 ph\u00e2n t\u00edch d\u1eef li\u1ec7u li\u00ean quan, b\u00e0i vi\u1ebft Data Scientist<\/a> s\u1ebd gi\u00fap l\u00e0m r\u00f5 s\u1ef1 kh\u00e1c bi\u1ec7t gi\u1eefa hai v\u1ecb tr\u00ed th\u01b0\u1eddng \u0111\u01b0\u1ee3c so s\u00e1nh n\u00e0y v\u00e0 h\u01b0\u1edbng n\u00e0o ph\u00f9 h\u1ee3p v\u1edbi \u0111\u1ecbnh h\u01b0\u1edbng c\u1ee7a b\u1ea1n h\u01a1n.<\/p>\n

\nG\u1ee3i \u00fd h\u1ecdc nhanh cho ng\u01b0\u1eddi m\u1edbi:<\/strong>\n

\u2013 B\u1eaft \u0111\u1ea7u v\u1edbi kh\u00f3a h\u1ecdc mi\u1ec5n ph\u00ed ChatGPT Prompt Engineering for Developers<\/em> c\u1ee7a deeplearning.ai (Andrew Ng) \u2014 2 gi\u1edd, t\u1eadp trung th\u1ef1c h\u00e0nh ngay v\u00e0o k\u1ef9 thu\u1eadt few-shot v\u00e0 Chain-of-Thought.<\/p>\n

\u2013 Th\u1ef1c h\u00e0nh h\u00e0ng ng\u00e0y tr\u00ean OpenAI Playground ho\u1eb7c Google AI Studio \u0111\u1ec3 r\u00e8n ph\u1ea3n x\u1ea1 \u0111i\u1ec1u ch\u1ec9nh prompt v\u00e0 \u0111\u1ecdc output c\u00f3 ph\u00ea ph\u00e1n.<\/p>\n

\u2013 Tham gia c\u00e1c group AI Vi\u1ec7t Nam tr\u00ean Facebook (AI Vietnam, AI\/ML Vietnam) \u0111\u1ec3 c\u1eadp nh\u1eadt JD th\u1ef1c t\u1ebf, trao \u0111\u1ed5i kinh nghi\u1ec7m v\u00e0 t\u00ecm c\u01a1 h\u1ed9i mentorship.<\/p>\n<\/div>\n

4. M\u1ee9c l\u01b0\u01a1ng Prompt Engineer t\u1ea1i Vi\u1ec7t Nam v\u00e0 qu\u1ed1c t\u1ebf n\u0103m 2025<\/h2>\n

\u0110\u00e2y l\u00e0 v\u1ecb tr\u00ed c\u00f3 m\u1ee9c thu nh\u1eadp bi\u1ebfn \u0111\u1ed9ng l\u1edbn do khan hi\u1ebfm nh\u00e2n s\u1ef1. Anthropic v\u00e0 OpenAI t\u1eebng \u0111\u0103ng tuy\u1ec3n Prompt Engineer v\u1edbi m\u1ee9c \u0111\u1ebfn $335,000\/n\u0103m cho c\u1ea5p Principal t\u1ea1i th\u1ecb tr\u01b0\u1eddng M\u1ef9. T\u1ea1i Vi\u1ec7t Nam, \u0111\u00e3 xu\u1ea5t hi\u1ec7n JD tr\u1ea3 80\u201390 tri\u1ec7u \u0111\u1ed3ng\/th\u00e1ng cho chuy\u00ean gia RAG v\u00e0 LangChain t\u1eeb 3 n\u0103m tr\u1edf l\u00ean t\u1ea1i c\u00e1c c\u00f4ng ty ph\u1ee5c v\u1ee5 th\u1ecb tr\u01b0\u1eddng qu\u1ed1c t\u1ebf.<\/p>\n\n\n\n\n\n\n\n\n
C\u1ea5p \u0111\u1ed9<\/th>\nKinh nghi\u1ec7m<\/th>\nL\u01b0\u01a1ng t\u1ea1i Vi\u1ec7t Nam (USD\/th\u00e1ng)<\/th>\nL\u01b0\u01a1ng to\u00e0n c\u1ea7u (USD\/n\u0103m)<\/th>\n<\/tr>\n<\/thead>\n
Junior Prompt Engineer<\/td>\n0\u20131 n\u0103m<\/td>\n$1,200\u2013$1,800<\/td>\n$75,000\u2013$120,000<\/td>\n<\/tr>\n
Mid-level Prompt Engineer<\/td>\n1\u20133 n\u0103m<\/td>\n$1,800\u2013$3,000<\/td>\n$120,000\u2013$200,000<\/td>\n<\/tr>\n
Senior Prompt Engineer<\/td>\n3\u20135 n\u0103m<\/td>\n$3,000\u2013$4,000<\/td>\n$200,000\u2013$300,000<\/td>\n<\/tr>\n
Lead \/ Principal<\/td>\n5+ n\u0103m<\/td>\n$4,000+<\/td>\n$300,000+<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

1. C\u00f4ng ty Vi\u1ec7t Nam \u0111ang tuy\u1ec3n Prompt Engineer<\/h3>\n

Nhu c\u1ea7u tuy\u1ec3n d\u1ee5ng trong n\u01b0\u1edbc t\u1eadp trung ch\u1ee7 y\u1ebfu \u1edf nh\u00f3m c\u00f4ng ty c\u00f4ng ngh\u1ec7, t\u00e0i ch\u00ednh v\u00e0 startup AI:<\/p>\n

\u2013 VinAI Research<\/strong> tuy\u1ec3n AI Engineer ki\u00eam Prompt Engineer cho d\u1ef1 \u00e1n LLM trong y t\u1ebf v\u00e0 xe t\u1ef1 h\u00e0nh \u2014 y\u00eau c\u1ea7u Python v\u00e0 kinh nghi\u1ec7m fine-tuning m\u00f4 h\u00ecnh.<\/p>\n

\u2013 Techcombank AI Lab<\/strong> c\u1ea7n Conversational AI Specialist x\u00e2y d\u1ef1ng chatbot ng\u00e2n h\u00e0ng tr\u00ean GPT-4 \u2014 \u01b0u ti\u00ean \u1ee9ng vi\u00ean c\u00f3 RAG v\u00e0 prompt safety.<\/p>\n

\u2013 FPT AI Center<\/strong> tuy\u1ec3n Prompt Engineer cho d\u1ef1 \u00e1n Document Intelligence, thi\u1ebft k\u1ebf pipeline t\u00f3m t\u1eaft v\u0103n b\u1ea3n ph\u00e1p l\u00fd v\u00e0 h\u1ee3p \u0111\u1ed3ng t\u1ef1 \u0111\u1ed9ng.<\/p>\n

\u2013 KMS Technology<\/strong> tuy\u1ec3n AI Solutions Engineer t\u00edch h\u1ee3p LLM v\u00e0o ph\u1ea7n m\u1ec1m xu\u1ea5t th\u1ecb tr\u01b0\u1eddng M\u1ef9 v\u00e0 \u00dac \u2014 y\u00eau c\u1ea7u LangChain v\u00e0 API integration.<\/p>\n

\u2013 Base.vn<\/strong> c\u1ea7n ng\u01b0\u1eddi am hi\u1ec3u LangChain v\u00e0 vector database \u0111\u1ec3 x\u00e2y d\u1ef1ng AI Assistant cho n\u1ec1n t\u1ea3ng qu\u1ea3n l\u00fd doanh nghi\u1ec7p n\u1ed9i \u0111\u1ecba.<\/p>\n

5. L\u1ed9 tr\u00ecnh tr\u1edf th\u00e0nh Prompt Engineer t\u1eeb \u0111\u1ea7u<\/h2>\n

Prompt Engineer kh\u00f4ng \u0111\u00f2i h\u1ecfi b\u1eb1ng c\u1ea5p chuy\u00ean bi\u1ec7t, t\u1ea1o ra c\u1ea3 c\u01a1 h\u1ed9i l\u1eabn th\u00e1ch th\u1ee9c do thi\u1ebfu khung \u0111\u00e0o t\u1ea1o chu\u1ea9n. L\u1ed9 tr\u00ecnh d\u01b0\u1edbi \u0111\u00e2y d\u1ef1a tr\u00ean JD th\u1ef1c t\u1ebf t\u1eeb c\u00f4ng ty Vi\u1ec7t Nam v\u00e0 qu\u1ed1c t\u1ebf, h\u01b0\u1edbng \u0111\u1ebfn m\u1ee5c ti\u00eau x\u00e2y d\u1ef1ng portfolio \u0111\u1ec3 \u1ee9ng tuy\u1ec3n v\u1ecb tr\u00ed Junior trong 6\u201312 th\u00e1ng h\u1ecdc nghi\u00eam t\u00fac.<\/p>\n

\nL\u01b0u \u00fd quan tr\u1ecdng tr\u01b0\u1edbc khi b\u1eaft \u0111\u1ea7u:<\/strong>\n

\u2013 L\u1ed9 tr\u00ecnh n\u00e0y gi\u1ea3 \u0111\u1ecbnh b\u1ea1n \u0111\u00e3 c\u00f3 Python c\u01a1 b\u1ea3n. N\u1ebfu ch\u01b0a bi\u1ebft Python, d\u00e0nh 2\u20133 th\u00e1ng \u0111\u1ea7u h\u1ecdc Python tr\u01b0\u1edbc khi chuy\u1ec3n sang b\u01b0\u1edbc ti\u1ebfp theo.<\/p>\n

\u2013 Th\u1ecb tr\u01b0\u1eddng AI thay \u0111\u1ed5i nhanh \u2014 t\u1eadp trung h\u1ecdc nguy\u00ean l\u00fd (CoT, RAG, evaluation framework) h\u01a1n l\u00e0 c\u00f4ng c\u1ee5 c\u1ee5 th\u1ec3 v\u00ec c\u00f4ng c\u1ee5 thay \u0111\u1ed5i li\u00ean t\u1ee5c.<\/p>\n

\u2013 Portfolio th\u1ef1c t\u1ebf quan tr\u1ecdng h\u01a1n ch\u1ee9ng ch\u1ec9 \u2014 nh\u00e0 tuy\u1ec3n d\u1ee5ng trong n\u01b0\u1edbc l\u1eabn qu\u1ed1c t\u1ebf \u0111\u1ec1u \u01b0u ti\u00ean d\u1ef1 \u00e1n c\u00f3 th\u1ec3 demo v\u00e0 gi\u1ea3i th\u00edch \u0111\u01b0\u1ee3c k\u1ebft qu\u1ea3.<\/p>\n<\/div>\n

1. Giai \u0111o\u1ea1n n\u1ec1n t\u1ea3ng (Th\u00e1ng 1\u20132)<\/h3>\n

M\u1ee5c ti\u00eau l\u00e0 hi\u1ec3u LLM ho\u1ea1t \u0111\u1ed9ng nh\u01b0 th\u1ebf n\u00e0o v\u00e0 th\u1ef1c h\u00e0nh vi\u1ebft prompt c\u00f3 \u00fd th\u1ee9c thay v\u00ec th\u1eed ng\u1eabu nhi\u00ean. H\u1ecdc kh\u00f3a ChatGPT Prompt Engineering for Developers<\/em> c\u1ee7a deeplearning.ai (mi\u1ec5n ph\u00ed, Andrew Ng) \u0111\u1ec3 n\u1eafm zero-shot, few-shot v\u00e0 Chain-of-Thought, \u0111\u1ed3ng th\u1eddi th\u1ef1c h\u00e0nh g\u1ecdi OpenAI API b\u1eb1ng Python v\u00e0 vi\u1ebft 50 prompt so s\u00e1nh k\u1ebft qu\u1ea3 c\u00f3 h\u1ec7 th\u1ed1ng.<\/p>\n

2. Giai \u0111o\u1ea1n k\u1ef9 n\u0103ng trung c\u1ea5p (Th\u00e1ng 3\u20135)<\/h3>\n

Giai \u0111o\u1ea1n n\u00e0y t\u1eadp trung LangChain, LlamaIndex v\u00e0 h\u1ec7 th\u1ed1ng RAG \u0111\u1ea7u ti\u00ean: n\u1ea1p PDF v\u00e0o vector database (Chroma ho\u1eb7c FAISS), x\u00e2y query pipeline \u0111\u1ec3 LLM tr\u1ea3 l\u1eddi t\u1eeb d\u1eef li\u1ec7u n\u1ed9i b\u1ed9 \u2014 k\u1ef9 n\u0103ng ph\u1ed5 bi\u1ebfn nh\u1ea5t trong JD Prompt Engineer t\u1ea1i Vi\u1ec7t Nam. Song song, h\u1ecdc prompt evaluation v\u1edbi ROUGE score v\u00e0 ph\u01b0\u01a1ng ph\u00e1p LLM-as-a-judge \u0111\u1ec3 \u0111\u00e1nh gi\u00e1 ch\u1ea5t l\u01b0\u1ee3ng output kh\u00e1ch quan.<\/p>\n

3. Giai \u0111o\u1ea1n ch\u1ee9ng ch\u1ec9 v\u00e0 portfolio (Th\u00e1ng 6\u201312)<\/h3>\n

X\u00e2y d\u1ef1ng 2\u20133 d\u1ef1 \u00e1n portfolio c\u00f3 th\u1ec3 demo: chatbot RAG t\u01b0 v\u1ea5n ph\u00e1p lu\u1eadt, ph\u00e2n t\u00edch h\u1ee3p \u0111\u1ed3ng t\u1ef1 \u0111\u1ed9ng ho\u1eb7c t\u00f3m t\u1eaft v\u0103n b\u1ea3n ti\u1ebfng Vi\u1ec7t v\u1edbi LLM. \u0110\u0103ng code l\u00ean GitHub \u2014 \u0111\u00e2y l\u00e0 “CV k\u1ef9 thu\u1eadt” th\u1ef1c s\u1ef1 v\u1edbi nh\u00e0 tuy\u1ec3n d\u1ee5ng AI. L\u1ea5y \u00edt nh\u1ea5t m\u1ed9t ch\u1ee9ng ch\u1ec9 uy t\u00edn: AWS Machine Learning Specialty<\/em>, Google Cloud Professional ML Engineer<\/em> ho\u1eb7c deeplearning.ai LangChain for LLM Application Development<\/em>.<\/p>\n

6. Prompt Engineer so v\u1edbi AI Engineer, ML Engineer v\u00e0 Software Engineer<\/h2>\n

Ranh gi\u1edbi gi\u1eefa c\u00e1c vai tr\u00f2 AI d\u1ea7n m\u1edd h\u01a1n khi c\u00f4ng ngh\u1ec7 ph\u00e1t tri\u1ec3n, nh\u01b0ng v\u1eabn c\u00f3 ph\u00e2n bi\u1ec7t r\u00f5 v\u1ec1 chi\u1ec1u s\u00e2u k\u1ef9 thu\u1eadt v\u00e0 ph\u1ea1m vi tr\u00e1ch nhi\u1ec7m. Hi\u1ec3u \u0111\u00fang t\u1eebng vai tr\u00f2 gi\u00fap x\u00e1c \u0111\u1ecbnh h\u01b0\u1edbng ph\u00f9 h\u1ee3p \u2014 v\u00e0 nhi\u1ec1u c\u00f4ng ty Vi\u1ec7t Nam hi\u1ec7n \u0111ang t\u00ecm ng\u01b0\u1eddi k\u1ebft h\u1ee3p \u0111\u01b0\u1ee3c c\u1ea3 prompt l\u1eabn l\u1eadp tr\u00ecnh.<\/p>\n\n\n\n\n\n\n\n\n\n
Ti\u00eau ch\u00ed<\/th>\nPrompt Engineer<\/th>\nAI Engineer<\/th>\nML Engineer<\/th>\nSoftware Engineer<\/th>\n<\/tr>\n<\/thead>\n
T\u1eadp trung ch\u00ednh<\/strong><\/td>\nThi\u1ebft k\u1ebf input\/output cho LLM<\/td>\nX\u00e2y d\u1ef1ng h\u1ec7 th\u1ed1ng AI \u0111\u1ea7u cu\u1ed1i<\/td>\nHu\u1ea5n luy\u1ec7n v\u00e0 t\u1ed1i \u01b0u m\u00f4 h\u00ecnh<\/td>\nPh\u00e1t tri\u1ec3n ph\u1ea7n m\u1ec1m t\u1ed5ng qu\u00e1t<\/td>\n<\/tr>\n
K\u1ef9 thu\u1eadt c\u1ed1t l\u00f5i<\/strong><\/td>\nPython, API, LangChain, RAG<\/td>\nPython, MLOps, cloud, CI\/CD<\/td>\nPyTorch, TensorFlow, to\u00e1n ML<\/td>\nBackend\/Frontend, databases<\/td>\n<\/tr>\n
To\u00e1n h\u1ecdc y\u00eau c\u1ea7u<\/strong><\/td>\nTh\u1ea5p<\/td>\nTrung b\u00ecnh<\/td>\nCao (th\u1ed1ng k\u00ea, \u0111\u1ea1i s\u1ed1 tuy\u1ebfn t\u00ednh)<\/td>\nTh\u1ea5p\u2013trung b\u00ecnh<\/td>\n<\/tr>\n
Th\u1eddi gian v\u00e0o ngh\u1ec1<\/strong><\/td>\n6\u201312 th\u00e1ng t\u1ef1 h\u1ecdc c\u00f3 h\u1ec7 th\u1ed1ng<\/td>\nCS\/IT + 1\u20132 n\u0103m kinh nghi\u1ec7m AI<\/td>\nTo\u00e1n\/CS + nghi\u00ean c\u1ee9u chuy\u00ean s\u00e2u<\/td>\n4 n\u0103m \u0111\u1ea1i h\u1ecdc ho\u1eb7c bootcamp<\/td>\n<\/tr>\n
L\u01b0\u01a1ng VN (mid-level)<\/strong><\/td>\n$1,800\u2013$3,000\/th\u00e1ng<\/td>\n$2,500\u2013$4,500\/th\u00e1ng<\/td>\n$3,000\u2013$6,000\/th\u00e1ng<\/td>\n$1,200\u2013$3,000\/th\u00e1ng<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

7. C\u00e2u h\u1ecfi th\u01b0\u1eddng g\u1eb7p<\/h2>\n

1. H\u1ecdc ng\u00e0nh g\u00ec \u0111\u1ec3 tr\u1edf th\u00e0nh Prompt Engineer?<\/em><\/strong><\/p>\n

Kh\u00f4ng c\u00f3 ng\u00e0nh h\u1ecdc d\u00e0nh ri\u00eang. Prompt Engineer \u0111\u1ebfn t\u1eeb nhi\u1ec1u n\u1ec1n t\u1ea3ng: l\u1eadp tr\u00ecnh vi\u00ean, ng\u00f4n ng\u1eef h\u1ecdc, vi\u1ebft n\u1ed9i dung, th\u1eadm ch\u00ed lu\u1eadt s\u01b0. B\u1eb1ng Computer Science, Linguistics ho\u1eb7c Data Science \u0111\u1ec1u l\u00e0 n\u1ec1n t\u1ea3ng t\u1ed1t nh\u01b0ng kh\u00f4ng b\u1eaft bu\u1ed9c \u2014 portfolio k\u1ef9 thu\u1eadt v\u00e0 t\u01b0 duy l\u1eadp lu\u1eadn c\u00f3 c\u1ea5u tr\u00fac quan tr\u1ecdng h\u01a1n t\u00ean ng\u00e0nh tr\u00ean b\u1eb1ng c\u1ea5p.<\/p>\n

2. Prompt Engineer c\u00f3 ph\u1ea3i xu h\u01b0\u1edbng nh\u1ea5t th\u1eddi kh\u00f4ng?<\/em><\/strong><\/p>\n

M\u1ed9t s\u1ed1 chuy\u00ean gia cho r\u1eb1ng khi AI ng\u00e0y c\u00e0ng th\u00f4ng minh, nhu c\u1ea7u thi\u1ebft k\u1ebf prompt s\u1ebd gi\u1ea3m. Tuy nhi\u00ean, WEF x\u1ebfp AI Prompt Engineering v\u00e0o danh s\u00e1ch k\u1ef9 n\u0103ng c\u1ea7n thi\u1ebft \u0111\u1ebfn n\u0103m 2027, v\u00e0 th\u1ef1c t\u1ebf ngay c\u1ea3 khi AI t\u1ef1 c\u1ea3i thi\u1ec7n, v\u1eabn c\u1ea7n ng\u01b0\u1eddi \u0111\u00e1nh gi\u00e1 ch\u1ea5t l\u01b0\u1ee3ng output v\u00e0 ki\u1ec3m so\u00e1t r\u1ee7i ro n\u1ed9i dung \u2014 nh\u1eefng k\u1ef9 n\u0103ng n\u00e0y s\u1ebd chuy\u1ec3n h\u00f3a th\u00e0nh vai tr\u00f2 AI Product ho\u1eb7c AI Operations ch\u1ee9 kh\u00f4ng bi\u1ebfn m\u1ea5t.<\/p>\n

3. C\u1ea7n bao nhi\u00eau th\u1eddi gian \u0111\u1ec3 b\u1eaft \u0111\u1ea7u \u1ee9ng tuy\u1ec3n v\u1ecb tr\u00ed Junior Prompt Engineer?<\/em><\/strong><\/p>\n

Ng\u01b0\u1eddi c\u00f3 n\u1ec1n t\u1ea3ng l\u1eadp tr\u00ecnh c\u1ea7n kho\u1ea3ng 4\u20136 th\u00e1ng h\u1ecdc nghi\u00eam t\u00fac (15\u201320 gi\u1edd\/tu\u1ea7n) \u0111\u1ec3 c\u00f3 portfolio \u1ee9ng tuy\u1ec3n Junior; ch\u01b0a bi\u1ebft Python th\u00ec c\u1ea7n th\u00eam 2\u20133 th\u00e1ng, t\u1ed5ng 8\u201312 th\u00e1ng. Y\u1ebfu t\u1ed1 quy\u1ebft \u0111\u1ecbnh l\u00e0 ch\u1ea5t l\u01b0\u1ee3ng portfolio \u2014 hai d\u1ef1 \u00e1n RAG c\u00f3 th\u1ec3 demo s\u1ebd thuy\u1ebft ph\u1ee5c nh\u00e0 tuy\u1ec3n d\u1ee5ng h\u01a1n nhi\u1ec1u ch\u1ee9ng ch\u1ec9 tr\u1ef1c tuy\u1ebfn c\u1ed9ng l\u1ea1i.<\/p>\n

Th\u1ecb tr\u01b0\u1eddng AI Vi\u1ec7t Nam \u0111ang b\u00f9ng n\u1ed5 v\u00e0 nhu c\u1ea7u nh\u00e2n s\u1ef1 Prompt Engineer<\/strong> s\u1ebd t\u0103ng \u00edt nh\u1ea5t 3\u20135 n\u0103m t\u1edbi, \u0111\u01b0\u1ee3c h\u1eadu thu\u1eabn b\u1edfi Quy\u1ebft \u0111\u1ecbnh 127\/Q\u0110-TTg v\u1ec1 Chi\u1ebfn l\u01b0\u1ee3c AI Qu\u1ed1c gia. \u0110\u00e2y l\u00e0 th\u1eddi \u0111i\u1ec3m l\u00fd t\u01b0\u1edfng \u0111\u1ec3 b\u1eaft \u0111\u1ea7u \u2014 kh\u00f4ng ph\u1ea3i v\u00ec ngh\u1ec1 \u0111ang “hot” m\u00e0 v\u00ec \u0111\u00e2y l\u00e0 b\u1ed9 k\u1ef9 n\u0103ng hi\u1ebfm c\u00f3, \u1ee9ng d\u1ee5ng r\u1ed9ng v\u00e0 kh\u00f3 t\u1ef1 \u0111\u1ed9ng h\u00f3a trong t\u01b0\u01a1ng lai g\u1ea7n.<\/p>\n

N\u1ed9i dung b\u00e0i vi\u1ebft mang t\u00ednh tham kh\u1ea3o. M\u1ee9c l\u01b0\u01a1ng v\u00e0 y\u00eau c\u1ea7u tuy\u1ec3n d\u1ee5ng thay \u0111\u1ed5i t\u00f9y theo doanh nghi\u1ec7p v\u00e0 th\u1eddi \u0111i\u1ec3m.<\/em><\/p>\n

Minh An<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"

Trong b\u1ed1i c\u1ea3nh AI b\u00f9ng n\u1ed5 to\u00e0n c\u1ea7u, th\u1ecb tr\u01b0\u1eddng lao \u0111\u1ed9ng xu\u1ea5t hi\u1ec7n v\u1ecb tr\u00ed ch\u01b0a t\u1eebng c\u00f3 trong b\u1ea3ng m\u00f4 t\u1ea3 ngh\u1ec1 nghi\u1ec7p …<\/p>\n","protected":false},"author":58,"featured_media":24847,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[17],"tags":[],"class_list":["post-24849","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tu-van-nghe-nghiep"],"_links":{"self":[{"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/posts\/24849","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/users\/58"}],"replies":[{"embeddable":true,"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/comments?post=24849"}],"version-history":[{"count":2,"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/posts\/24849\/revisions"}],"predecessor-version":[{"id":25738,"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/posts\/24849\/revisions\/25738"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/media\/24847"}],"wp:attachment":[{"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/media?parent=24849"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/categories?post=24849"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.careerlink.vn\/cam-nang-viec-lam\/wp-json\/wp\/v2\/tags?post=24849"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}