Skip to content
View skalrn's full-sized avatar

Block or report skalrn

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
skalrn/README.md

Deepak Kollipalli

Engineering leader. 18 years building payments, trading, and back-office systems, and 5+ years leading distributed teams. Based in Vancouver, BC.

Work

  • Payments and money movement. Built a team of 4 from scratch at Arcesium (a D. E. Shaw spin-off) and led a wire transfer platform for hedge fund and asset management back offices, handling 50 to 100 wires a day up to about $20M per wire under same-day cut-offs. Architected its SWIFT messaging layer, so field-level MT103, MT202 and MT202COV construction, ACK/NACK handling, and MT900/MT910 reconciliation. Moved it from an internal application to multi-tenant SaaS on AWS with an amount-based approval rules engine.
  • Trading and back office. Eight years at D. E. Shaw, ending as project leader over API-based trade services and a flexible trade data model covering equities, repos, FX and swaps, giving back-office and finance teams one view for reconciliation, positions and P&L. Cut manual trade booking time about 6x.
  • E-commerce and identity. Headless commerce at Elastic Path for merchants including T-Mobile, The Pokemon Company and Charlotte Tilbury. Stripe payment integration, OpenID Connect SSO, and a cart promotions engine. Managed 3 engineers in an 11-person distributed team across Canada and Europe and cut sprint-goal failure 20%.

Side projects

  • purpl-brain: decision memory and a pre-action guardrail for codebases edited by multiple independent AI coding agents. An agent calls analyze_impact over MCP to see which prior decisions a change would break before it writes code. Neo4j decision graph plus Qdrant semantic search, so contradictions that share no wording are still caught. Write-up.
  • career-ops interview modes: plan, practice, and debrief modes plus the session-transcript producer, merged into santifer's career-ops. Built to its never-fabricate rule; other contributors' features now build on the transcript format.
  • Repo-context-aware AI code review: RAG over repository context (n8n, Qdrant, LLMs) that generates project-specific feedback on GitHub pull requests.

What I work with

Java, Spring Boot, JavaScript · distributed systems, microservices, event-driven · SWIFT MT, Stripe, OpenID Connect · PostgreSQL, MySQL, SQL Server · AWS (EC2, S3, Lambda, SQS), Docker, CI/CD · RAG, MCP servers, agent memory and guardrails · Neo4j, Qdrant, Claude Code, Ollama

AWS Certified Generative AI Developer (Professional) · AWS Certified Solutions Architect (Associate)

Elsewhere

Writing on Medium · CV and contact on LinkedIn

Popular repositories Loading

  1. repo-context-aware-rag-ai-code-review repo-context-aware-rag-ai-code-review Public

    Automate GitHub pull request reviews with n8n workflows using Retrieval-Augmented Generation (RAG), AI language models, and a Qdrant vector store. Build and maintain repo context embeddings to deli…

    2

  2. purpl_brain purpl_brain Public

    Shared decision memory for multi-agent codebases - contradiction detection, cross-session retrieval, MCP interface

    TypeScript

  3. career-ops career-ops Public

    Forked from santifer/career-ops

    AI-powered job search system built on Claude Code. 14 skill modes, Go dashboard, PDF generation, batch processing.

    JavaScript

  4. skalrn skalrn Public