Role & responsibilities

We are a Chennai-based technology group building an enterprise AI platform. We are hiring the engineer who will own it technically, end to end.

About the role

This is a hands-on technical leadership role. You will own the architecture, write and review code, and make the decisions that determine whether the platform actually works in production how retrieval is built, how quality is measured, what a request costs, and how the system stays secure.

Retrieval quality and evaluation are the most important engineering problems here, not an afterthought bolted on at the end. If you have shipped a RAG system and know precisely why it worked or why it did not, this role is aimed at you.

Tech stack

Python and TypeScript/Node • PostgreSQL with pgvector • OpenAI and Anthropic APIs in a multi-provider runtime • RAG (chunking, embeddings, hybrid search, reranking, context assembly) • document processing and extraction pipelines • AWS (ECS, S3, SQS, IAM) with infrastructure as code • Docker • REST and Server-Sent Events • evaluation harnesses and LLM observability

What you will do

- Own the end-to-end architecture, the technical roadmap, and the non-functional requirements — latency, cost per request, reliability and security
- Design and operate the multi-provider LLM runtime: routing, fallback, prompt and context management, streaming, cost control
- Build and continuously improve the retrieval pipeline, and own evaluation — build the harness, define the metrics, and gate releases on results rather than intuition
- Reduce hallucination through grounding and citation
- Build workers that parse, extract from and structure messy real-world documents at scale
- Deploy and operate on AWS with tracing, logging, cost dashboards and alerting
- Own the security surface: key management, secrets, encryption, access control and audit trails
- Hire and lead the AI engineering team, and set the standards for code review, testing and release




- Define how AI-assisted development tooling is used in the codebase, and own the quality gates around it

Skills and experience we are looking for

- 510 years in professional software engineering, including at least 2 years building LLM systems that ran in production with real users
- You have shipped and operated a retrieval-augmented generation system at meaningful scale a demo or proof of concept is not sufficient
- Strong backend engineering in Python and/or TypeScript and Node
- Solid PostgreSQL and hands-on vector search — pgvector, Qdrant, Weaviate, Milvus or equivalent
- Production cloud experience, preferably AWS: containers, queues, object storage, IAM
- Direct integration with model provider APIs (OpenAI, Anthropic), not solely through frameworks
- Evaluation discipline — you can explain concretely how you knew your system was right, and what you did when it was not
- Clear technical communication with non-engineers
- Able to work in person in Chennai; relocation support is available

Positive to have Open-source contributions to AI or ML projects (the strongest single signal for us). Inference optimization — vLLM, quantization, batching, caching. Agent architectures, tool use, Model Context Protocol. Security engineering or applied cryptography. Rust. Published research, a strong Kaggle record, or conference talks. Experience leading a small engineering team. Document AI, OCR pipelines or large-scale information extraction. Multilingual or Indic-language AI systems.

Other details

Compensation is benchmarked against funded AI companies, with equity discussed openly and early. The process is an intro call, a technical deep-dive on a system you have built, a system design exercise, and a conversation with the CEO — no HR screening round, and paid work if we ask for anything substantial in your own time. This is a confidential search; full details of the platform and the organization are shared under NDA at the first technical conversation.

To apply: [email protected]

📌 Head of AI / AI Technical Lead (Chennai)
🏢 Atman Artwork
📍 Chennai