Autonomy-first SWE · Go + distributed systems · LLM eval / RAG / agents · Freelance + mentoring
Start here: llm-tournament · order-demo · go-edge-mesh
Open knowledge / archive: modern-swe-library · Contact: mailto:cariyaputta@gmail.com · https://www.linkedin.com/in/lavantien/
Featured work (dense table)
| Area | Repo | What it shows | Why it matters |
|---|---|---|---|
| LLM evaluation | llm-tournament |
eval harness + benchmarking platform | reproducibility, cost/quality tradeoffs, real tooling |
| Backend service | order-demo |
production-style Go service demo | API design, auth, testing, deployability |
| Systems / mesh | go-edge-mesh |
distributed data processing system | systems thinking, integration + ops concerns |
| Distributed systems | distributed-booking-microservices |
gRPC + NGINX + OpenAPI | service boundaries, reliability patterns |
| Enterprise full‑stack (OG stack) | spring-angular-ecom-architecture |
Spring Boot + Angular e‑commerce architecture (multi‑module, hand‑wired build/CI patterns) | “manual” enterprise engineering: wiring, modules, integration patterns (legacy but real-world) |
| Dev environment | dotfiles |
Neovim/WezTerm workflow | craft + reproducible setup |
What I do (consulting / freelancing)
- Backend & APIs: Go, REST/gRPC, Postgres, performance, concurrency
- Distributed systems: boundaries, observability, reliability patterns
- LLM engineering: eval design, datasets/metrics, RAG/agent pipelines, cost control
- Mentoring: architecture reviews, refactors, onboarding, “ship safely, ship often”
“Vibe engineering” (what I mean)
Fast iteration with guardrails:
- small milestones + visible progress
- tight feedback loops (demos, PRs, benchmarks)
- tests where it matters; metrics where it hurts
- clear interfaces; fewer moving parts
- docs that explain tradeoffs (“why”), not just steps (“how”)
Minimal ceremony, not minimal rigor.
Working agreements (how I operate)
- Short kickoff → clarify constraints + success metrics
- 1–2 page design note → align before building
- Small PRs → easy to review, easy to stop/redirect
- Weekly demo/check-in (or async updates) → no surprises
- Respect access boundaries + security policies
- No reusing client code/data across clients
Values + boundaries (intentional filter)
I’m explicitly pro–open knowledge and anti-gatekeeping.
Open-first (preferred):
- I love shipping in the open (OSS, public docs, reproducible builds).
- If it must be closed-source, I still aim for: portability, open standards, low lock-in.
Hard no:
- spyware/surveillance, coercive monitoring
- scams, plagiarism, piracy in client work
- DRM / anti-user lock-in as the main product goal