▊

    

      GitHub
      Email
      LinkedIn
      Instagram
      
    

  


  

    

      

        
        
        
        vikram@obsidian — zsh — type help
      

      

      

        vikram@obsidian:~$
        
      

    

  


  

    
ABOUT // whoami

    

      I build idealistic AI agents that solve real-world problems — retrieval
      pipelines that cite their sources, agents that plan, and vision models that see. 199 repositories deep,
      and every one is a lesson. I don't stop at a notebook: I ship the full stack, and I care about systems
      that prove what they claim.
    

    

    GitHub stats: repositories, stars, followers, contributions, streaks
  


  

    
LIVE TELEMETRY // regenerated daily from real GitHub data

    

      No shields.io. No third parties. Every pixel below is rendered by a GitHub Action inside this repo.
    


    

      
 AGENT LOG // tail -f agent.out

      

    


    

    Isometric 3D contribution city — every tower is a real day of code
    

    Rolling 56-day contribution heatmap — one new frame every day
    

    

      Self-hosted terminal boot sequence
      Stats flipbook — contributions, streaks, repos, top language, mission
    

    

      Language donut chart by real bytes of code
      Live public activity stream
      GitHub Wrapped year in code
      Contribution galaxy — one star per commit
      Repository traffic and latest build status
      Repository constellation map
    

  


  

    
FEATURED SYSTEMS // shipped

    Retrieval augmented generation pipeline
    

      
        
🔍 citecheck

        
Research agent where every citation is programmatically verified. Hand-rolled tool loop, guardrails at the capability layer, eval suite that publishes its own failures. Runs on a 6 GB GPU with no API key.

        
python · guardrails · evals

      
      
        
🕸️ LangGraph-Agent

        
Multi-agent research workflow: analyst personas, parallel interview loops with Tavily web search, and cited report synthesis.

        
langgraph · tavily · agents

      
      
        
⚡ production-rag

        
Event-driven RAG pipeline on Inngest workflows, FastAPI function host, OpenAI embeddings and Qdrant vector search.

        
inngest · fastapi · qdrant

      
      
        
🖼️ colpali-rag

        
Multimodal RAG with ColPali late-interaction visual retrieval and hybrid dense + BM25 search over Qdrant, with retrieval evals.

        
colpali · hybrid search · evals

      
      
        
⛏️ MOIL-Manganese-AI

        
AI + space-tech decision-support platform for MOIL Ltd — maps reserves, forecasts production, predicts shortfall risk (SIH 2026).

        
forecasting · geospatial · impact

      
      
        
📈 alpaka-mcp

        
MCP server + CLI for the Alpaca Trading API — trade stocks, ETFs and crypto through AI assistants and the terminal.

        
mcp · cli · trading

      
    


    
PROJECT EXPLORER // live from data.json

    

    

      
Loading projects…

    

  


  

    
GUESTBOOK // sign it, I read every one

    Guestbook wall — signatures from the pinned issue
    

      Reaction meter — reactions on the pinned issue
      

        
✍️ Add your name

        
Comment on the pinned guestbook issue and you appear on the wall above within a day. Hit a reaction to move the meter. Be honest — would you hire me?

        
→ Open the guestbook issue

      

    

  


  

    
STACK // real deps

    Tech stack
    

      PythonLangChainLangGraph
      OpenAIHugging FaceColPali
      FAISSQdrantChromaDB
      PyTorchTensorFlowscikit-learn
      OpenCVMediaPipeFastAPI
      StreamlitGradioReact
      pandasNumPyDocker
    

  


  

    
NOTES // field reports

    
    
→ All notes  ·  RSS

  


  

    
CONTACT // ping me

    

      ✉️ vikrambtech2025@gmail.com  · 
      💼 LinkedIn  · 
      📸 Instagram  · 
      🐙 GitHub
    

    

      Machine-readable profile: data.json · resume.json ·
      résumé · API-style endpoint, no auth.