Languages
- Python
- C++
- C
- JavaScript
- R
Bengaluru, India — CSE @ RVCE · Class of 2028
Surendra Vishwakarma
Full-Stack & Applied-AI Developer
I build full-stack products and AI-driven systems end to end — React dashboards, FastAPI backends, RAG pipelines and security tooling for real problems.
Computer Science undergraduate at R. V. College of Engineering (CGPA 9.05). I like taking hard problems — LLMs over messy financial PDFs, anomaly detection on industrial networks — and turning them into working, shippable systems.
- 9.05CGPA · RVCE B.E. CSE
- 120+DSA problems solved
- 6public GitHub projects
Open to software engineering internships
- Full-Stack Systems
- Applied AI & RAG
- ICS/SCADA Security
- MERN Stack
- PyTorch & LLMs
- Real-Time Web
- Open Source
01 — About
Where I am, and what I'm building toward
I'm a Computer Science Engineering undergraduate at R. V. College of Engineering, Bengaluru, currently in my fifth semester with a 9.05 CGPA. My foundation is core CS — data structures, OOP, operating systems, networks and system design — backed by 120+ solved DSA problems and comfortable C/C++ implementations.
The problems I enjoy most sit where products meet infrastructure: making LLMs reliable over messy real-world PDFs, spotting anomalies in industrial control traffic, and shipping full-stack platforms with real-time features. I gravitate toward systems that are honest about failure — test-driven agents, sandboxed execution, explainable ML rather than black boxes.
I learn by building in public. Six projects now span full-stack (MERN, Socket.IO, OAuth), applied AI (RAG pipelines, LSTM autoencoders, recommender models) and security (honeypots, STIX threat-intel sharing). I prefer keeping data local and on-device where it makes sense, and instrumenting systems so improvements are measurable instead of vibes-based.
02 — Skills
What I work with
AI & Machine Learning
- PyTorch
- scikit-learn
- RAG
- LLM Integration
- sentence-transformers
- LSTM Anomaly Detection
- AI Agents
- MCP
- Ollama
Web & Frameworks
- React.js
- Node.js
- Express.js
- Flask
- FastAPI
- Celery
- Socket.IO
- Tailwind CSS
- MERN Stack
Databases
- MongoDB
- PostgreSQL
- ChromaDB
- Redis
Security & Networking
- ICS/SCADA Honeypots
- Modbus TCP
- STIX 2.1 / TAXII 2.1
- MITRE ATT&CK
- JWT & OAuth
CS Fundamentals & Tools
- Data Structures & Algorithms
- OOP
- Operating Systems
- Computer Networks
- System Design
- Git
- GitHub
- Docker
- Postman
03 — Featured Projects
Systems I've designed and shipped
Cybersecurity · Applied AI
Shadow-OT — ICS/SCADA Deception Platform
A deception and threat-intelligence platform for industrial control networks: honeypots that mirror real PLCs, ML-based anomaly detection, and automated containment playbooks.
- 15 Docker Compose services — Modbus TCP, SSH and SMB honeypots mirroring real PLCs
- PyTorch LSTM autoencoder flags anomalous OT traffic by reconstruction error
- SOAR playbook engine for automated isolation and alerting, plus an APT profiler
- STIX 2.1 / TAXII 2.1 threat-intel server and a red-team Modbus attack simulator
- Flask + Socket.IO API with a React dashboard and auto-generated PDF incident reports
- Docker Compose
- Modbus TCP
- Flask
- Socket.IO
- React
- PyTorch
- STIX 2.1
- MITRE ATT&CK
LLM · RAG · Local-First
Annual Report Analysis System
A fully local annual-report analyzer: LLMs run on-device via Ollama behind an OpenAI-compatible client, turning PDFs into structured KPIs, risk flags and cited chat answers — financial data never leaves the machine.
- Two-pass extraction of 28+ financial KPIs with a multi-domain risk classifier — financial, regulatory, cyber, climate and supply-chain
- Hybrid RAG Q&A with conversational memory and page-level source citations
- Footnote & consistency engine, year-over-year anomaly detection and embedding-based peer discovery
- Cut PDF detection time from 10–15s to 3–5s (~3x) and reduced LLM token usage by 60%
- FastAPI
- Celery
- Redis
- React
- Ollama
- ChromaDB
- sentence-transformers
- PyMuPDF
Full-Stack · Explainable ML
SkillSwap — Skill Exchange & Mentorship Platform
A MERN platform for skill exchange and mentorship with real-time sessions and an explainable ML matchmaking layer.
- Socket.IO real-time session chat, Jitsi video sessions, Google/GitHub OAuth plus JWT authentication
- Explainable ML layer: linear-model skill recommender and mentor ranking, TF-IDF cosine-similarity matchmaking, logistic-regression session-success prediction
- OTP-verified college emails, gamification, learning roadmaps and an analytics dashboard
- MongoDB
- Express
- React
- Node.js
- Socket.IO
- Jitsi
- JWT
- TF-IDF
LLM · Safety
In progress
Autonomous Coding Agent with Sandboxed Execution
An autonomous coding agent with a write → run → observe → fix loop that executes LLM-generated code inside an isolated Docker sandbox.
- Test-driven execution so every generated change must prove itself
- AST-based guardrail layer that blocks unsafe code before it executes
- Every attempt instrumented to report task success and safety-block rates
- Docker
- LLM
- pytest
Repository link coming soon
04 — Education & Achievements
Academic track record
-
2024 — 2028
R. V. College of Engineering
B.E. in Computer Science & Engineering · Currently 5th semester · CGPA 9.05
Core: Data Structures & Algorithms, OOP, Operating Systems, Computer Networks, System Design
-
2022 — 2024
Vishwachetana Vidyaniketa Residential PU College
12th (PCMB) · 95.67%
JEE 95.67 percentile · CET rank 840 · COMEDK rank 1648
-
2022
Sri Sharana Basaveshwara English Medium High School, Karatagi
10th grade · 99.2%
achievements
- 120+DSA problems solved across coding platforms, building strong problem-solving fundamentals
- Meta PyTorch OpenEnv HackathonParticipated in the hackathon
- 6 public projectson GitHub, spanning full-stack, AI/ML and cybersecurity
05 — Contact
Let's build something real.
I'm seeking a software engineering internship and I'm always up for technical collaboration — a project, a hackathon, or an interesting problem. Use the form below or reach out directly; every message lands in my inbox.