Turn your AI & software vision into reliable, production‑grade reality.

I’m a full‑stack developer with deep focus on building practical LLM‑powered RAG Agents.
Early on, I spent lots of time doing traditional software development: building back‑ends, front‑ends, APIs and deploying real‑world projects. As generative AI grew popular, I noticed a common pain point: many AI prototypes look impressive in demos but break badly in real use — poor retrieval accuracy, serious hallucinations, unstable agent tool calling, and cannot connect to real business data.
That pushed me to dive deep into RAG and agent engineering. Instead of only chasing fancy demos, I focus on building systems that actually work for production: tuning chunking strategies, optimizing vector retrieval, designing reliable agent workflows, mitigating hallucination, and integrating with existing business systems.
I still keep my full‑stack roots. When an AI project needs custom backend logic, frontend interfaces, container deployment or debugging, I can handle the whole stack without bringing in extra developers.
For me, it is never just writing code. My goal is to understand your real‑world requirements, avoid over‑engineered fancy solutions, and deliver maintainable, deployable software and AI agents that truly solve your problems. I value transparent communication, clear progress updates and clean, well‑documented deliverables.

Work Terms

Scope Definition
All deliverables, features and requirements must be clearly documented at project start. Any new features, changed business logic, additional data sources or new agent capabilities after kick‑off count as out‑of‑scope work and will require a separate quote and timeline adjustment.
Revisions
Each project includes 2 rounds of consolidated revisions per milestone. A revision means adjustments & bug‑fixes within the agreed‑upon scope. Brand‑new features, completely changed logic or new workflow designs are treated as scope changes, not revisions. Extra revisions will be charged at my hourly rate.
Client Responsibilities
Client shall provide required documents, API keys, credentials, sample data and feedback within reasonable business time. Project timelines will be extended accordingly if client‑side input is delayed.
Payment & Milestones
Projects run by milestones via Guru escrow. Work only starts after corresponding milestone funds are secured. Final source‑code handover happens upon full milestone release. No full source preview before payment completion.
Intellectual Property
Full source‑code & project IP transfers to client only after full milestone payments are completed. Pre‑existing general‑purpose code snippets, utility libraries and open‑source components remain my or open‑community property. I reserve rights to showcase anonymized project work in my portfolio.
Support & Warranty
7‑day bug‑fix warranty is included after final delivery, only for bugs inside the agreed‑scope deliverables. Warranty does not cover new features, environment migration, third‑party API/vector‑service outages, model‑provider rate‑limit issues or self‑inflicted misconfiguration after handover. Ongoing maintenance is available as extra retainer service.
Cancellation
If client cancels mid‑project, payment is due for all completed work. Unfinished future milestones will be cancelled.