Sourish Chakraborty
Senior Data Engineer · Ørsted
Kuala Lumpur, Malaysia · 25 years in tech
Professional Summary
I build AI systems and data platforms — from raw infrastructure to the LLM integration layer that sits on top. 25 years across GIS, enterprise software, cloud-native architecture, and applied AI, currently at Ørsted in Kuala Lumpur.
My focus is on the hard operational problems: data pipelines that don't drift, LLM deployments that don't leak PII, Kubernetes clusters that don't page at 3am. I care deeply about clean code, observable systems, and the craft of building things that actually stay built.
Outside work I build open-source tools — Dhi (FIM autocomplete for VS Code) and LLM Governance Engine (PII firewall for LLMs) — and write about the lessons learned building them.
Core Competencies
AI Systems & LLM Engineering
LLM integration, PII governance, FIM autocomplete, RAG pipelines, MLOps infrastructure.
Data Platform Engineering
High-throughput data pipelines, lakehouse architecture, real-time and batch processing.
Cloud & Platform Engineering
AWS, Azure, Kubernetes, IaC, microservices, scalable cloud-native architecture.
DevOps & Site Reliability
CI/CD automation, observability, incident response, resilience engineering.
Experience
Featured Writing
I Built a PII Firewall for LLMs in a Weekend (and Caught My Own Leak)
How I built an open-source PII firewall and governance policy engine for LLMs in a weekend — blocks sensitive data before it reaches GPT-4o, with YAML rules, audit trail, and webhook alerts.
How Dhi Understands Your Entire Codebase: Multi-Language Tree-sitter, BM25, and Hybrid Search
How to add full repository intelligence to an AI IDE: multi-language Tree-sitter chunking (Go, Rust, Java, JavaScript), BM25 hybrid search with RRF, .gitignore-aware file discovery, and new /index-dir and /search API endpoints.
FIM Autocomplete at < 150ms: Tree-sitter, nomic Embeddings, and StarCoder2-3B on a Laptop
Step-by-step: build a < 150ms fill-in-the-middle autocomplete engine using Tree-sitter chunking, nomic-embed-text, Chroma, and StarCoder2-3B. Part 1 of building Dhi — the open-source AI coding IDE.
Open Source
Career Highlights
- ▸Increased high-frequency data processing capacity by 90%+ at Ørsted.
- ▸Led multiple legacy-to-cloud platform migrations for mission-critical workloads.
- ▸Designed and delivered CI/CD automation across cloud and on-prem environments.
- ▸Built Kubernetes platforms supporting mission-critical microservices at scale.
- ▸Created LLM-integrated data pipelines that now process production workloads.
Certifications
Tools & Technologies
Cloud
AWS, Azure, Kubernetes, EKS, AKS
IaC
Terraform, CloudFormation, Ansible, Packer
CI/CD
Azure DevOps, GitHub Actions, AWS CodePipeline
Observability
Prometheus, Grafana, ELK, Zabbix
Languages
Python, Go, Bash, PowerShell, TypeScript
AI / LLM
FastAPI, Ollama, OpenAI, Presidio, LangGraph
Languages
Open to consulting
AI infrastructure, data platform engineering, and cloud architecture. Reach me at sourish1978@gmail.com