I'm a researcher-engineer with a sharp focus on LLMs, multi-modal AI systems, and intelligent infrastructure for verticalized domains like finance, health, and industrial systems. I specialize in fine-tuning foundation models, building graph-augmented RAG pipelines, and deploying production-grade AI systems with full observability and modularity.
- 🔍 LLM Fine-Tuning: Specialized instruction-tuned models (LLaMA 2, Falcon) for financial analysis, compliance, macroeconomics & private markets
- ⚙️ LLM Ops: End-to-end training pipelines with
QLoRA,MLflow,Prefect,Ray,WandB, and vector DBs likeWeaviateandQdrant - 🧱 RAG + Knowledge Graphs: Embedding-backed retrieval with
FAISS, hybrid scoring, and graph-based reasoning (Neo4j + TransE) - 🧑⚕️ Domain AI: Building LLM agents & workflows for auditors, doctors, and strategists using structured + unstructured data
- 🌐 Infra + Control Planes: Multi-tenant orchestration using Docker, NATS JetStream, S3, Temporal, Rancher and Centrifugo
Python · TypeScript · Go · Rust · PyTorch · Transformers · HuggingFace
TGI · vLLM · FastAPI · SvelteKit · Next.js · Ray · Prefect · BentoML
S3 · Trino · Iceberg · delta-rs · Polars · Docker · NATS · Temporal
AWS · GCP · Azure · Postgres · Redis · Weaviate · Qdrant . Langchain
LlamaIndex · LangGraph . Langfuse . LLM · RAG · OpenAPI · REST · WebSockets
- Fine-tuning + evaluating LLMs for reasoning over private financial data
- Designing RAG vs. fine-tuning benchmarking systems for explainability & cost
- Architecting LLM agents for document QA + task flows (on-chain, enterprise, or clinical)
- 🧪 Website — AI Infrastructure, Smart Manufacturing, & Vertical SaaS
- 📩 Open to research collabs, early-stage builds, or just geeking out on graphs + LLMs