Anuj dev Singh
AI Engineer. LLMs, RAG, agents, building intelligence
New to Contra
Anuj dev is ready for their next project!
oss-veda — AI-Powered Open Source Discovery Engine
What it does
oss-veda is a Claude Code plugin that helps developers discover, evaluate, and contribute to open source projects using native AI subagents. Instead of manually browsing GitHub, you just describe what you're looking for and oss-veda does the rest.
Technical Architecture
3 Skills: Discovery, Evaluation, Contribution mapping
5 Native Subagents: Each specialized for a different research task
2 Slash Commands: /discover and /evaluate
Distribution: Published via anuj-ai-tools marketplace repo
Runtime: uv + native Claude Code subagents
Key Highlights
🤖 Built on Claude Code's native subagent architecture
🔍 Semantic OSS discovery beyond keyword search
⚡ Distributed as an installable Claude Code plugin
🛠️ Full agentic workflow — search, evaluate, recommend in one command
📦 Published to marketplace for other developers to use
Built For
Developers who want AI to handle OSS research so they can focus on building.
Stack: Python · Claude Code · uv · GitHub API · Anthropic API
#ClaudeCode #AIAgents #OpenSource #Python #LLM #AIEngineering #MultiAgent #DeveloperTools #AnthropicAPI #GitHub
1
36
BirdCLEF+ 2026 — Kaggle Bronze Medal 🥉
Achievement
Top 9% out of 4,084 teams — Bronze Medal on Kaggle's BirdCLEF+ 2026 competition. Mean CV AUC of 0.9564.
What it does
Built a deep learning pipeline to identify bird species from audio recordings — a real-world bioacoustics problem used by wildlife conservationists and researchers globally.
Technical Approach
Model: EfficientNet-B0 ensemble trained on mel-spectrograms
Data Pipeline: Audio → mel-spectrogram conversion → augmentation → classification
Ensemble Strategy: Multiple model checkpoints averaged for robust predictions
Validation: Cross-validated AUC of 0.9564 across species classes
Key Highlights
🥉 Kaggle Bronze Medal — top 9% of 4,084 teams
🎯 0.9564 mean CV AUC
🔊 End-to-end audio ML pipeline from raw .ogg files
📊 EfficientNet-B0 with spectrogram-based vision approach
🌿 Real-world conservation application
Skills:
Computer Vision · Audio ML · Deep Learning · PyTorch · Signal Processing · Model Ensembling
Stack: Python · PyTorch · EfficientNet · Librosa · Kaggle
#DeepLearning #ComputerVision #AudioML #PyTorch #Kaggle #MachineLearning #EfficientNet #SignalProcessing #Python #AIEngineering
0
35
Engram — Self-Hosted Company Knowledge Brain
What it does
Engram is an MCP (Model Context Protocol) server that turns your company's scattered knowledge into a unified, queryable brain. It ingests data from Slack, Notion, Gmail, and GitHub — and makes all of it instantly accessible to AI assistants like Claude.
The Problem it Solves
Engineers waste hours hunting through Slack threads, Notion pages, and old PRs just to answer "how did we decide this?" Engram fixes that. Ask once, get the answer instantly — with full source attribution.
Technical Architecture
Vector Store: PostgreSQL + pgvector for semantic search
Graph Layer: Kuzu graph database for relationship mapping between documents and people
MCP Server: Native integration with Claude and other AI assistants
Ingestion Pipelines: Slack, Notion, Gmail, GitHub connectors
Self-hosted: Full data privacy, runs on your own infrastructure
Key Features
🔍 Semantic search across all connected sources
🕸️ Graph-based relationship mapping (who said what, linked to which doc)
🔒 Fully self-hosted — your data never leaves your infra
⚡ Real-time ingestion from live workspace tools
🤖 Plug-and-play with Claude via MCP protocol
Built For
Startups and engineering teams that want AI that actually knows their codebase, decisions, and context — not generic answers.
#RAG #LLM #AIEngineering #EnterpriseAI #MCP
0
42
FinGenius — AI Personal Finance Advisor
Upload your real bank statement → get an instant dashboard and chat with an AI advisor grounded in your actual spending.
FinGenius turns a raw bank statement into an instant financial dashboard and a conversational AI advisor that answers from your real numbers — "where's my money going?", "what's my biggest expense?", "where can I cut back?"
It started in 2024 as a Gemini-powered Colab notebook exploring five Gen AI capabilities. I rebuilt it into a deployable full-stack app: real bank-statement parsing (CSV/Excel, debit/credit splits, day-first dates, UPI merchant extraction, INR-aware), a live dashboard (KPIs, cash-flow, savings rate, 50/30/20 budget, investment calculator), and an always-on AI chat rail.
Under the hood, a LangGraph ReAct agent orchestrates all five Gen AI capabilities as tools — structured output (Pydantic), RAG over a financial knowledge base (Chroma), embeddings + semantic search over your transactions, and five financial calculators.
Built for public demo: per-user session isolation (256-bit HttpOnly cookies), PII redaction before anything reaches the LLM, in-memory-only statements with auto-expiry, plus rate limiting and a daily spend cap.
Python · FastAPI · LangChain · LangGraph · OpenAI GPT-4o-mini · ChromaDB · RAG · Embeddings · pandas
Links to attach
GitHub: https://github.com/anujdevsingh/financial_genius_agent
(https://github.com/anujdevsingh/financial_genius_agent)Portfolio: https://www.codewithanuj.com/
0
52