Hello, I build things that I imagine.
nagendhra@ai:~$ cat /etc/profile
AI Engineer with 4+ years building enterprise-grade generative AI systems, retrieval-augmented generation pipelines, and agentic architectures. I ship production systems that serve thousands of users — not just prototypes.
- Previously Data Engineer at Goldman Sachs, where I validated 57M+ financial records with zero incidents
- MS in Information Science (AI & Data Analytics) from University at Albany, SUNY — GPA: 3.70/4.0
- Currently building agentic RAG systems and open-source developer tools in NYC
|
Cloning OpenClaw (the open-source personal AI assistant) but rebuilding it to work entirely without LLM API calls. No tokens. No API keys. No cloud dependency. The approach: Instead of sending prompts to an LLM, users teach the system directly. You show it patterns, correct its behavior, and build up its knowledge through a learning-by-teaching method — like training a junior developer by pair programming, not by paying OpenAI. |
How it works:
Traditional OpenClaw My Version
┌─────────────────┐ ┌─────────────────┐
│ User asks query │ │ User asks query │
│ ↓ │ │ ↓ │
│ Send to LLM API │ │ Match against │
│ ↓ │ │ taught patterns │
│ Wait for response │ │ ↓ │
│ ↓ │ │ User corrects │
│ Pay $$$ tokens │ │ if wrong │
│ ↓ │ │ ↓ │
│ Hope it's right │ │ System learns │
└─────────────────┘ │ permanently │
└─────────────────┘
Cost: $0 | Latency: ~0ms
Core idea: Every correction makes it smarter. Every interaction is training data. No API calls, no rate limits, no vendor lock-in. Your assistant runs on your machine, learns from your workflow, and belongs to you.
|
Persistent Memory for AI Coding Tools
Branch-aware embeddings + ChromaDB vector storage. Cuts token usage by 60-80% per session. Extends context retention across the full AI lifecycle. |
Self-Healing AI Runtime Engine
Detects failures, traces root causes across cascading crashes, auto-recovers production systems. 58 Go packages, 3 SDKs, single binary. |
|
Agentic RAG Job Intelligence Platform
Aggregates roles from 17K+ company career pages with autonomous LangGraph agents. Real-time job discovery for thousands of users. |
Real-Time Financial Analytics & Anomaly Detection
Ingests product events, models KPIs, detects anomalies with ML scoring. Kafka + DuckDB + dbt + scikit-learn pipeline. |
+---------------------------+-----------------------------+-----------------+
| ROLE | COMPANY | PERIOD |
+---------------------------+-----------------------------+-----------------+
| AI Systems Engineer | Self-Employed (NYC) | Jan 2026 - Now |
| Graduate Research Asst. | University at Albany (SUNY) | Mar - Dec 2025 |
| Data Engineer | Goldman Sachs (Hyderabad) | Jan 2022 - 2023 |
+---------------------------+-----------------------------+-----------------+
