atQor
Website:
atqor.com
Job details:
Job Title: Tech Lead - AI
Location: Ahmedabad (Work from Office)
What You'll Do
- Design and build AI agents — single-agent, multi-agent, and human-in-the-loop systems for enterprise use cases.
- Architect end-to-end GenAI solutions — RAG, agentic RAG, tool-using agents, structured outputs, evaluation pipelines, and guardrails.
- Own the technical reference architecture for AI engagements — model selection, orchestration framework, memory, retrieval, observability, and cost optimization.
- Write production code. Lead by example in the IDE — agent loops, tool definitions, prompt scaffolding, evaluation harnesses, and integration layers.
- Lead code and design reviews across the AI team; raise the bar on quality, testability, and reliability.
- Drive prompt engineering and evaluation as engineering disciplines — versioning, regression testing, offline/online eval, and quality metrics.
- Mentor AI engineers and developers transitioning into agentic and GenAI work.
- Partner with pre-sales on solution design, POCs, and technical demos for strategic deals.
- Stay on the frontier — evaluate new models, frameworks, and patterns; bring the right ones into atQor's stack.
What You Bring
- 6–12 years in software engineering, with at least 3 years building production GenAI / agentic systems.
- Deep hands-on experience — this is not a managerial role; you will be in code reviews and on the keyboard.
- Strong system design skills — can take an ambiguous customer problem to a working agent in days, not months.
- Excellent technical communication — can explain a multi-agent architecture to a CTO and a junior engineer in the same meeting.
- Bias for shipping — comfortable with the trade-offs of getting reliable AI into production.
Core Technical Skills (Must-Have)
Languages & Engineering Foundations
- Python (primary) — strong proficiency in modern Python, async patterns, type hints, packaging, testing.
- TypeScript / Node.js — for agent UIs, tool servers, and Microsoft 365 / Copilot extensibility.
- Solid grounding in REST/GraphQL APIs, async messaging, microservices, and event-driven architectures.
- Comfort with Git, CI/CD, containerization (Docker), and IaC (Bicep / Terraform).
LLMs & Foundation Models
- Hands-on with Azure OpenAI, OpenAI, Anthropic Claude, and open-source models (Llama, Mistral, Phi).
- Deep understanding of tokenization, context windows, function/tool calling, structured outputs, streaming, and caching.
- Experience with fine-tuning, distillation, and model evaluation (helpful, not mandatory).
Agent Frameworks & Orchestration
- Production experience with at least two of: LangChain / LangGraph, Semantic Kernel, AutoGen, CrewAI, Microsoft Agent Framework, Pydantic AI.
- Hands-on with Microsoft 365 Copilot extensibility — declarative agents, custom engine agents, Copilot Studio.
- Experience designing multi-agent systems — supervisor/worker, planner/executor, debate, and tool-routing patterns.
- Familiarity with MCP (Model Context Protocol) and modern tool/function-calling standards.
Retrieval & Knowledge Systems
- Strong RAG fundamentals — chunking, embeddings, hybrid search, reranking, query rewriting.
- Hands-on with Azure AI Search, vector DBs (pgvector, Pinecone, Weaviate, Qdrant), and graph-based retrieval.
- Experience with Microsoft Fabric / OneLake, SharePoint, and Microsoft Graph as enterprise knowledge sources.
Evaluation, Safety & Observability
- Built and run evaluation pipelines — golden sets, LLM-as-judge, regression testing, A/B evaluation.
- Familiar with tools like Azure AI Foundry evaluations, LangSmith, Langfuse, Promptfoo, Ragas.
- Working knowledge of AI safety, prompt injection defenses, PII handling, content filtering, and responsible AI practices.
Cloud & Platform
- Strong Microsoft Azure depth — App Service, Functions, AKS, API Management, Key Vault, Cosmos DB, Service Bus.
- Familiarity with Azure AI Foundry, Azure ML, and AI Hub for model lifecycle management.
- DevOps for AI — prompt versioning, model deployment, telemetry, cost monitoring, and rollback strategies.
Click on Apply to know more.