Experience
Software Engineer
Aug 2024 – Present
- Built Enceladus, a full-stack app that turns Mitsubishi GxWorks3 HTML uploads into AI-generated documentation with a follow-up chat — owning the full surface from PostgreSQL schema and FastAPI backend to the React/TypeScript frontend.
- Engineered a provider-agnostic, multi-interface AI agent system with a three-stage generate → respond → validate pipeline (Azure, OpenAI, Anthropic, Gemini, GitHub Copilot, Ollama, and OpenAI-compatible endpoints).
- Led a structured LLM evaluation study benchmarking 8+ models on PLC code comprehension; drove an agentic dev workflow (OpenSpec + GitHub Copilot) to accelerate delivery.
- Shipped a parsing/transformation engine converting GxWorks3 PLC programs into IEC 61131-10 standardized XML & Structured Text as a framework-level release.
Associate Software Engineer
Jul 2023 – Jul 2024
- Automated 700+ manual test cases with Robot Framework and Python, building a reusable automation library and Boofuzz fuzz-testing scripts that lifted coverage and product robustness.
- Delivered two proof-of-concept projects (ALT-WAN, Enceladus) on deadline, mentored 10+ peers, and set up local Ollama LLM dev environments for the team.
Trainee Engineer (Internship)
Jan 2023 – Jul 2023
- Built a Python GUI app for the Automation Department, a packaging tool bundling Python libraries into a portable executable, and an automated Outlook email-reply system (Python + win32com).
Open-source projects
Cascade-AIShipped
Multi-provider agent CLI that routes tasks across Anthropic, OpenAI, Gemini, Azure, and Ollama with failover, permission escalation, and per-tier budget control — plus a live agent-topology dashboard.
TypeScript · Node.js · React · Ink · MCP · npm (MIT)mrxIn dev
Multi-role model orchestration CLI that assigns LLMs to distinct roles (reasoner / executor / tool-caller) in one session, pairing a strong reasoner with a fast executor to cut inference cost.
TypeScript · Node.js · Multi-provider LLMMurmurIn dev
Privacy-first, always-on reminder engine with a rolling 30s in-RAM audio buffer transcribed fully on-device via whisper.cpp — no audio ever leaves the phone.
Flutter · Dart · Kotlin · On-device MLMore on GitHub
Industrial tooling, automation libraries, and experiments — plus the projects above with full source.
github.com/Varun-SV →Technical skills
Languages
Frameworks & libraries
AI / ML
DevOps & tools
Databases & domain