class NikhilAI:
designation = "AI Product Developer Intern @ Rooman Technologies"
mission = "Ship production-grade AI — not demos, not notebooks."
builds = ["LLM Pipelines", "Voice Agents", "RAG Systems", "AI SaaS"]
active_build = "GhostFolio — AI Startup Revival Intelligence Platform"
targeting = ["AI Engineer", "Software Engineer", "AI/SWE Intern roles"]
def status(self): return "🔴 ONLINE · Bangalore, India"|
AI Startup Revival Intelligence Platform Ingests failed startup data → runs Gemini API analysis → outputs revival strategies, TAM scores, pivot vectors & viability ratings.
STATUS: |
Real-time Multi-Persona STT→LLM→TTS Streaming voice pipeline with live persona switching. Resolved async deadlocks, interruption handling & silent handoffs in production.
STATUS: |
|
AI Crime Pattern Prediction · IJRAR Published XGBoost ensemble on real Bengaluru crime datasets. 87% accuracy. FastAPI real-time inference layer.
STATUS: |
🟢 NPM Hawknpm Package Intelligence Monitor Search, inspect & analyze any npm package in real-time — version history, download stats, dependencies, bundle size & security signals. Built for developers who don't want to tab-hop between registries.
STATUS: |
|
Developer Debugging Tool Intelligent error analysis and fix suggestions for developers.
STATUS: |
⚡ LLM & Generative AI
| Domain | Technologies |
|---|---|
| Foundation Models | GPT-4, Gemini 1.5, Claude 3.5, Mistral, LLaMA 3, Phi-3 |
| Core Techniques | Prompt Engineering, Chain-of-Thought, Few-Shot, Zero-Shot, Tree-of-Thought |
| Retrieval & Memory | RAG, Embeddings, Vector Search, Semantic Retrieval, FAISS, ChromaDB |
| Agent Systems | AI Agents, Multi-Agent Orchestration, Function Calling, Tool Use, ReAct |
| Inference Infra | HuggingFace Transformers, Ollama (Local LLM), vLLM concepts |
| Fine-tuning | LoRA, QLoRA, PEFT, Quantization (GGUF/GPTQ concepts) |
| Evaluation | LLM Benchmarking, BLEU/ROUGE, Human Eval, Hallucination Detection |
🎙️ Voice & Real-time AI
| Domain | Technologies |
|---|---|
| STT | Whisper, Deepgram, Real-time streaming STT |
| TTS | ElevenLabs, Kokoro, Coqui, Bark — latency benchmarking |
| Pipeline | STT → LLM → TTS, Async Streaming, WebSocket orchestration |
| Solved in Prod | Async Deadlocks, Interruption Handling, Session Resets, Persona Switching |
🧠 Machine Learning & Data Science
| Domain | Technologies |
|---|---|
| Frameworks | Scikit-learn, XGBoost, LightGBM, CatBoost |
| Deep Learning | PyTorch (concepts), TensorFlow (concepts), Neural Networks |
| NLP | Tokenization, Embeddings, Transformers, Sentiment Analysis, NER |
| Data | Pandas, NumPy, Matplotlib, Seaborn, Plotly |
| Techniques | Feature Engineering, Ensemble Methods, Cross-Validation, Hyperparameter Tuning |
| Evaluation | AUC-ROC, F1, Precision/Recall, Confusion Matrix, SHAP |
🔧 Backend & Infrastructure
| Domain | Technologies |
|---|---|
| Languages | Python · SQL · TypeScript |
| Frameworks | Django · FastAPI · React · TailwindCSS |
| Databases | PostgreSQL · Supabase · Redis · SQLite |
| DevOps | Docker · Git · Vercel · Azure |
| AI Tooling | n8n · Cursor AI · Claude Code · GitHub Copilot |
[AZ-900] Microsoft Certified: Azure Fundamentals
[INF-AI] Infosys: Principles of Generative AI
[ANT-101] Anthropic: Claude 101
[ANT-CC] Anthropic: Claude Code in Action
[IJRAR] Research Publication: AI-Based Crime Pattern Prediction
TARGETING: AI Engineer · SWE · AI/SWE Intern
nikhilchowdary334@gmail.com · Bangalore, India · Open to relocation