I'm B V S Aneesh — an AI engineer and full-stack developer based in Visakhapatnam, India. I build intelligent systems and full-stack applications: from AI-powered platforms to data-rich products, with a core focus on engineering depth and real-world impact.
I love working at the edge of what's possible — RAG systems, multi-agent architectures, and intelligent products grounded in real data. I'm just as comfortable modelling a database and wiring up APIs as I am crafting the interface on top of them. My stack of choice is TypeScript/Next.js on the frontend and Python/FastAPI or NestJS on the backend, deployed cloud-native on Vercel and Docker.
Currently, I'm the founder of Dinomem — a Postgres-native memory layer and SDK for AI agents, featuring hybrid semantic + BM25 + entity-graph retrieval, Gemini re-ranking, CRDT conflict resolution, TTL, and BYOK encryption — shipping as @dinomem/sdk (TypeScript) and dinomem-py (Python). I've competed in 8 hackathons across India building autonomous agents, AI finance platforms, and agentic SaaS products.
Open To: Full-time SWE / AI Engineer roles · ML engineering internships · AI product collaborations · Open source contributions
| Domain | Proficiency | Details |
|---|---|---|
| LLM Engineering | ████████████ Expert | GPT-4 · Gemini · Groq / Llama 3 · Prompt engineering · Streaming inference |
| RAG Pipelines | ████████████ Expert | LangChain · LangGraph · pgvector · Hybrid retrieval · Document ingestion |
| Autonomous Agents | ███████████░ Advanced | Multi-agent orchestration · Supervisor loops · Tool use · Self-healing CI |
| Memory Systems | ███████████░ Advanced | Postgres-native memory · Semantic + BM25 + entity-graph · CRDT resolution |
| AI Product Design | ███████████░ Advanced | Streaming UX · Multi-tenant SaaS · Human-in-the-loop · Agentic workflows |
| NLP & Embeddings | ██████████░░ Advanced | Semantic search · Gemini embeddings · Chunking strategies · Re-ranking |
| MLOps & Infra | █████████░░░ Intermediate | Docker orchestration · Webhook pipelines · Sandboxed execution environments |
| Data Science | █████████░░░ Intermediate | NumPy · Pandas · Matplotlib · Exploratory analysis · Data visualisation |
🧠 Dinomem — Postgres-Native Memory Layer for AI Agents
Dinomem is a production-grade memory infrastructure for AI agents. It provides hybrid retrieval combining semantic vector search, BM25 full-text matching, and an entity-relationship graph, with Gemini-powered re-ranking on top. The system ships as two first-class SDKs — @dinomem/sdk for TypeScript and dinomem-py for Python — making it drop-in compatible with any LLM stack. Features include CRDT conflict resolution for concurrent agent writes, TTL-based memory expiry, and BYOK encryption for enterprise deployments. Currently in active development — public release coming soon.
| Attribute | Details |
|---|---|
| Stack | TypeScript · Python · Postgres · pgvector · Hono · Gemini |
| Scale | Multi-agent concurrent writes · CRDT conflict resolution · TTL lifecycle management |
| Performance | Hybrid retrieval: semantic + BM25 + entity-graph · Gemini re-ranking |
| Security | BYOK encryption · Tenant-scoped memory isolation |
| Impact | Drop-in memory SDK for any AI agent · Eliminates stateless agent limitations |
| Organisation | Dinomem |
💬 Chatly — Multi-Tenant SaaS RAG Workspace
Chatly is an AI-powered multi-tenant SaaS platform that allows organisations to ingest private knowledge bases and query them through a real-time streaming RAG chat interface. The architecture enforces hard tenant isolation so no data leaks across organisations. It features three-tier RBAC (owner, admin, member) and Auth0 Organizations for enterprise B2B authentication. The RAG pipeline is built on LangGraph with MongoDB Atlas as the document store. Currently in active development — launching soon.
| Attribute | Details |
|---|---|
| Stack | NestJS · MongoDB Atlas · Auth0 · LangGraph · TypeScript |
| Scale | Multi-tenant architecture · Hard tenant data isolation · Three-tier RBAC |
| Performance | Streaming RAG chat interface · Real-time token streaming |
| Security | Auth0 Organizations · B2B enterprise auth · Tenant-scoped data access |
| Impact | Private knowledge base chat for teams · Enterprise-grade SaaS architecture |
| Repository | rooney011/chatly-saas |
💰 Fincil Remastered — AI Financial Council Platform
Fincil Remastered is a rebuilt Next.js 15 AI finance platform where a multi-agent AI Council debates and negotiates purchase decisions in real time via live streaming. The platform layers deterministic financial mathematics under the LLM reasoning layer, uses Supabase for auth and database, and applies pgvector-based retrieval-augmented generation to surface relevant financial context during the decision loop. Built and demoed across multiple hackathons including Google TechSprint at GITAM University and Hogwarts Hackathon at GVP College.
| Attribute | Details |
|---|---|
| Stack | Next.js 15 · Supabase · Vercel AI SDK · pgvector · TypeScript |
| Scale | Real-time multi-agent streaming · pgvector-powered RAG retrieval |
| Performance | Live token streaming · Deterministic financial math layer |
| Security | Supabase auth · Row-level security · Scoped user data |
| Impact | AI-native personal finance · Hackathon-tested across two national events |
| Repository | rooney011/Fincil-RE |
⚙️ CodeWeaver — Autonomous AI SRE Incident Response Agent
CodeWeaver is an autonomous SRE agent that closes the entire incident response loop without human intervention. When a production alert fires via webhook, CodeWeaver reads application logs from shared Docker volumes, invokes a Groq-hosted LLM to diagnose the root cause, generates a Python remediation script, executes it against live service APIs, and validates that the system has returned to a healthy state — all within seconds. Ships with an integrated chaos-app for continuous chaos engineering and resilience validation. Built and demoed at the AI Buildathon by CrossStack.
| Attribute | Details |
|---|---|
| Stack | Python · FastAPI · Groq (Llama 3) · Docker · Docker Compose |
| Scale | Containerised microservices · Webhook-driven alerting · Shared log volumes |
| Performance | MTTR from minutes to seconds · Always-on autonomous response loop |
| Security | Sandboxed execution · Human-approval gate available · Scoped API access |
| Impact | Eliminates on-call burnout · Consistent incident resolution · Proactive chaos testing |
| Repository | rooney011/CodeWeaver |
🔧 AutoHeal-CI — Autonomous CI/CD Pipeline Healing Agent
AutoHeal-CI is a multi-provider autonomous agent that monitors CI/CD pipelines, identifies failing tests and broken code, generates minimal JSON patch diffs via LLM inference, validates each fix in an isolated sandbox, and pushes successful patches back to the repository without human input. Supports Groq (Llama 3), OpenAI (GPT-4), DeepSeek, and Gemini as provider backends. Features a real-time React dashboard with live streaming logs and fuzzy patch matching. Built and placed at RIFT '26 Hackathon, PW Institute of Innovation, Bangalore.
| Attribute | Details |
|---|---|
| Stack | Python · FastAPI · React · TypeScript · Groq · OpenAI · Gemini · DeepSeek |
| Scale | Handles Syntax, Logic, Dependency, and Indentation failure classes |
| Performance | Sub-second patch generation · Streaming dashboard · Automated regression tests |
| Security | Sandboxed fix validation · No direct push without passing test suite |
| Impact | Autonomous CI healing · Multi-LLM provider redundancy · Reduced developer toil |
| Repository | rooney011/AutoHeal-CI |
📊 APEX — Formula 1 Telemetry OS
APEX is a race-engineer's telemetry OS that turns raw Formula 1 lap data into a 3D track ribbon, scrub-synced D3 charts, and exportable generative lap fingerprints. Ships with ten hand-baked pole lap datasets spanning 2018-2025 and a live OpenF1 next-session countdown. The Python pipeline pre-bakes hero lap comparisons with smoothed speed traces for instant load on the Next.js frontend.
| Attribute | Details |
|---|---|
| Stack | Next.js · React Three Fiber · D3.js · FastF1 · Python · Vercel |
| Scale | Multi-season historical telemetry · Lap-granularity data resolution |
| Performance | Pre-baked hero laps · Smoothed rendering pipelines · Edge deployment |
| Security | Public read-only telemetry data · Static asset delivery |
| Impact | Motorsport intelligence accessible to fans and analysts without specialist tooling |
| Repository | rooney011/Apex |
ZigmaNeural · Remote · Feb 2026 – Present
Contributing to software development and AI initiatives — building and optimising tech solutions while collaborating with the team on production-level projects.
Grafx IT Solutions · Visakhapatnam · Oct 2024 – Dec 2024
Worked with Python and core data-science libraries across multiple dataset projects — performing data cleaning, analysis, and visualisation to extract meaningful insights.
| Recognition | Details |
|---|---|
| Founder — Dinomem | Building Postgres-native memory infrastructure for AI agents as an independent startup |
| 8 Hackathons | Competed across India: RIFT '26 · Innovators 2026 · Google TechSprint · AI Buildathon · Hogwarts · TECHNOVA-2025 · GenAI Hackathon · Vibe Coding |
| RIFT '26 — Bangalore | Built AutoHeal-CI: autonomous CI/CD healing agent at PW Institute of Innovation |
| Google TechSprint — GITAM | Built Fincil: multi-agent AI financial council with real-time negotiation loop |
| AI Buildathon — CrossStack | Built CodeWeaver: autonomous SRE incident response agent with LLM-driven healing |
| 299 Contributions | 299 GitHub contributions in the last year across production AI systems |
| GitHub PRO | Active GitHub PRO member · Consistent contribution streak |
| OSS Contributor | Contributing to oss-chatbot-with-rag, hive agent framework, and is-a.dev registry |
| Event | Date | Venue | Project |
|---|---|---|---|
| RIFT '26 | Feb 19-20, 2026 | PW Institute of Innovation, Bangalore | AutoHeal-CI — Autonomous CI/CD Healing Agent |
| Innovators Hackathon 2026 | Jan 22-23, 2026 | NMIET Campus, Pune | BrandMatch Pro — Creator Business OS |
| Vibe Coding Hackathon | Jan 5-6, 2026 | ANITS College, Visakhapatnam | Drift — AI Browser Extension for Focus |
| Google TechSprint | Jan 3-4, 2026 | GITAM University, Visakhapatnam | Fincil — Multi-Agent AI Financial Council |
| AI Buildathon by CrossStack | Dec 22-27, 2025 | a-hub, Visakhapatnam | CodeWeaver — Autonomous SRE Agent |
| Hogwarts Hackathon | Dec 6-7, 2025 | GVP College, Visakhapatnam | Fincil — AI Finance Debate Engine |
| TECHNOVA-2025 | Sep 19-20, 2025 | ANITS College, Visakhapatnam | Voice Payments App for Elderly Users |
| GenAI Hackathon | Sep 15-16, 2025 | Vignan's Institute, Visakhapatnam | PayLink — Shareable Payment Links |
current_focus:
building:
- Dinomem — Postgres-native memory layer & SDK for AI agents (startup)
- Chatly — Multi-tenant SaaS RAG workspace (launching soon)
learning:
- Advanced multi-agent orchestration with LangGraph
- CRDT algorithms and distributed memory systems
- LLM evaluation frameworks (LangSmith, Arize)
exploring:
- Agentic RAG with self-correction and reflective retrieval
- Serverless AI inference on edge networks
- AI observability tooling and benchmark design
open_to:
- Full-time SWE / AI Engineer roles (India or Remote)
- AI/ML engineering internships
- Open-source collaboration on AI agent frameworks
- Technical co-founding opportunities in the AI product space
"I build intelligent systems and full-stack applications — from AI-powered platforms to data-rich products — with a core focus on engineering depth and real-world impact."