> system.identity
AI engineer building LLM applications that are useful after the demo:
retrieval systems, tool-using agents, evaluation harnesses, and cloud backends.
> system.now
Associate AI Consultant @ Siemens DISW
I work at the intersection of agent behaviour, reliable retrieval, and production delivery. My recent work has involved taking enterprise AI copilots from ambiguous requirements to pilot deployment—designing the data paths, agent logic, evaluation strategy, and operational guardrails around them.
| Supported by a multi-user artifact-sync and RAG platform | Enterprise data indexed with tenant-aware retrieval workflows | Estimated task-time reduction through an agentic copilot |
flowchart LR
A["Natural-language requirement"] --> B["Context engineering"]
B --> C["Agent orchestration"]
C --> D["Tools + schemas"]
D --> E["Evaluation + observability"]
E --> F["Production workflow"]
style A fill:#0f172a,color:#fff,stroke:#334155
style B fill:#1e3a8a,color:#fff,stroke:#60a5fa
style C fill:#0e7490,color:#fff,stroke:#67e8f9
style D fill:#155e75,color:#fff,stroke:#67e8f9
style E fill:#164e63,color:#fff,stroke:#a5f3fc
style F fill:#065f46,color:#fff,stroke:#6ee7b7
Jul 2026 — present
Focus: Scaling the AI Copilot for Teamcenter Pipeline Designer
→ broader rollout
→ maintainable codebase
→ CI/CD with custom DeepEval suites
→ Arize-based observability
Jun 2025 — Jun 2026
Built: Multi-tenant RAG and artifact-sync platform on Azure
Owned: Chunking, rate limiting, auto-indexing, isolation, and pilot delivery
Designed: LangChain + LangGraph copilot translating natural-language requirements
into automation-ready pipeline actions
Dec 2024 — Jan 2025
Built: Flask + PostgreSQL APIs connected to Google Earth Engine
Explored: Structured LLM outputs and tool calling for geospatial analysis
Outcome: 30% performance improvement after modularising the codebase
|
A multi-agent productivity system that converts natural-language requests into Notion and email workflows.
|
An end-to-end Marathi speech-to-text pipeline fine-tuned from Wav2Vec2-BERT.
|
llm_applications:
- Azure OpenAI
- RAG architecture
- prompt and context engineering
- token-budget management
agentic_systems:
- LangChain
- LangGraph
- smolagents
- tool use and function calling
- MCP server development
retrieval_and_data:
- Azure AI Search
- hybrid semantic/vector search
- FAISS
- Qdrant
- PostgreSQL
- Cosmos DB
evaluation_and_operations:
- DeepEval
- Arize
- OpenTelemetry
- Application Insights
- KQL
- CI/CD
engineering:
- Python
- FastAPI
- Flask
- Docker
- Pydantic
- C++
- B.E. in Artificial Intelligence and Data Science, CGPA: 9.4
- Problem Statement Winner and Top 5 Finalist — ISRO Bhartiya Antariksh Hackathon 2024
- Top 5 Finalist — Citi Campus Innovation Challenge 5.0
- Best Paper Award — WCSC 2024, Data Analytics
- Scopus-indexed publication and Hacktoberfest 2024 contributor
Interested in building AI systems that are measurable, maintainable, and genuinely useful.
Start a conversation ·
View portfolio