LLMs & RAG · Python · FastAPI · Kubernetes · Terraform · AWS · Azure · DevSecOps
I build AI-driven products end to end — RAG pipelines, LLM-powered developer tooling, ML inference services — on top of 8+ years running cloud infrastructure at scale. That pairing is the whole point: prototyping with a model is the easy part, and everything after it (serving, securing, scaling, and paying for it) is where projects actually live or die.
🔭 Currently building — LLM-powered developer tooling: AI security triage that separates real risk from scanner noise (VulnSift) and a secrets platform that stops credentials before they reach a commit (SecretSentinel).
- LLM applications — RAG pipelines end to end (chunking, embeddings, vector search, citation-grounded answers), Claude & OpenAI API integration, prompt/cost controls, streaming UX
- ML services — reproducible training pipelines, real-time and batch inference APIs, versioned model artifacts with metrics, health probes and deployment readiness
- Platform & IaC — EKS / AKS, reusable Terraform module libraries, GitOps (ArgoCD), self-hosted & air-gapped LLM serving, self-service infrastructure
- DevSecOps & SRE — SAST / SCA / IaC scanning, SBOM and supply-chain security (SLSA), SLOs, observability (Prometheus / Grafana / OpenTelemetry), FinOps
| Project | What it is | Stack |
|---|---|---|
| RAG-DocQA | Document Q&A over PDF / DOCX / Markdown — semantic retrieval, streaming answers, every response traced back to its source chunk | FastAPI · ChromaDB · sentence-transformers · React |
| SalesCoPilot | RAG sales assistant — account intelligence over CRM data, personalized email generation, call-transcript summarization | LangChain · ChromaDB · GPT-4 Turbo · Streamlit |
| ResumeForge | Resume analysis & ATS optimization — parsing, skills-gap scoring against job posts, 7+ templates, DOCX/PDF export | FastAPI · SQLAlchemy · OpenAI · React |
| Customer-Retention-Intelligence-Platform | Churn-prediction service — reproducible training pipeline, single and batch (500-record) inference, liveness/readiness probes | scikit-learn · FastAPI · Docker |
| Project | What it is | Stack |
|---|---|---|
| VulnSift | AI security triage — Claude scores real-world risk and flags false positives across SARIF / Snyk / Semgrep / Trivy, then generates fix patches and PRs | Python · Claude API · GitHub Actions |
| SecretSentinel | Secrets security platform — pre-commit scanning CLI, AES-256-GCM vault, SQS-driven rotation workers, Prometheus metrics, React dashboard | Go · FastAPI · Node.js · PostgreSQL · React |
| pipeline-armor | Reusable DevSecOps GitHub Actions — SAST, container/IaC scanning, secret detection, SBOM, and one unified deploy gate | GitHub Actions · Python |
| terraform-aws-blueprint | Production-grade AWS module library — VPC · EKS · RDS · IAM · KMS · S3 — validated inputs, examples, CI and tfsec gating | Terraform |
AI/ML
Languages & Frameworks
Cloud & Containers
IaC, CI/CD & GitOps
Observability & Security
Microsoft AI & ML Engineering · IBM DevOps & Software Engineering · IBM Full Stack Developer · Google IT Support


