Project Role : Custom Software Engineer
Project Role Description : Develop custom software solutions to design, code, and enhance components across systems or applications. Use modern frameworks and agile practices to deliver scalable, high-performing solutions tailored to specific business needs.
Must have skills : Large Language Models (LLMs)
Good to have skills : NA
Minimum 7.5 year(s) of experience is required
Educational Qualification : 15 years full time education
SSummary:
As a Custom Software Engineer Lead, a typical day involves creating tailored software solutions by designing, coding, and improving various components within systems or applications. The role requires working with contemporary development frameworks and following agile methodologies to ensure the delivery of scalable and efficient software that meets unique business requirements. Collaboration with team members and adapting to evolving project needs are integral parts of the daily workflow, fostering innovation and continuous improvement in software products.
Roles & Responsibilities:
- Design and implement multi-agent orchestration pipelines using LangGraph state
machines — conditional routing, self-healing retry loops, and supervisor / sub-agent
hierarchies
- Engineer retrieval-augmented generation (RAG) systems: chunking, embedding
pipelines, vector search, re-ranking, and token-budget management
- Model knowledge graphs (entities, relationships, ontologies) and build hybrid
retrieval that fuses graph traversal with vector similarity
- Govern prompts through a versioned prompt registry implement multi-model
routing by task/stage with fallback strategies
- Build LLM evaluation harnesses — golden datasets, automated scoring, regression
gates, and hallucination/grounding checks
- Implement guardrails: schema-constrained (structured/JSON) output, validation,
self-correction, and safety filtering
- Serve and optimize on-prem LLMs via vLLM alongside cloud OpenAI-compatible
APIs — batching, streaming, context-window management
- Instrument the AI stack for observability — token usage, latency percentiles, cost
attribution, and trace/provenance logging
- Integrate agentic interoperability protocols (A2A, MCP) and connect agents to
enterprise tools via REST/GraphQL
Professional & Technical Skills:
Must To Have Skills:
Core AI & LLM
- Python 3.10+ — production async code, FastAPI,
- LLM engineering — prompt design, token budgeting, multi-model routing, structured
output, self-healing retry loops
- Agent frameworks — LangGraph / LangChain (state machines, tool calling,
orchestration)
- Model serving — vLLM, TGI, or OpenAI-compatible inference APIs (streaming,
batching)
Evaluation & Reliability
- LLM evaluation — eval datasets, automated scoring, regression testing,
hallucination/grounding metrics
- Guardrails — output validation, schema enforcement, safety filtering
- Observability — token/latency/cost metrics, tracing, provenance
Nice-to-Have Skills
-Ontology modeling — taxonomies and relationship schemas for domain knowledge
-Dual-tier agent memory — Redis (L1) + mem0 (L2), semantic recall, context
consolidation
- DAG workflow compilation — topological sorting, cycle detection, parallel execution
(Ray)
- A2A / MCP / AG-UI — emerging agentic interoperability protocols
- Fine-tuning / LoRA / quantization — model adaptation and optimization
- RAG architecture — chunking strategies, re-ranking, context assembly, grounding
- Vector databases — Qdrant, Pinecone, or Weaviate embedding pipelines and
semantic search
- Embeddings — sentence-transformers or equivalent similarity tuning
- Knowledge graphs — Neo4j / Cypher, entity-relationship modeling, graph traversal
Qualifications
- 7-8 years professional software engineering experience, with 3+ years building LLM,
agent, or ML systems in production ,5+ years on Python .
- Strong async Python and hands-on experience with an agent framework
(LangGraph, LangChain, or equivalent)
- Demonstrated experience with RAG,…