Enterprise AI product builder · Agentic systems architect · Digital transformation leader
I turn AI ideas into production systems — agent workflows, retrieval pipelines, developer tools, and SaaS infrastructure. Public GitHub is my open lab.
I sit at the intersection of product strategy, AI engineering, and enterprise transformation. Most public repositories here are deliberately small — they prove out one idea cleanly, with tests, documentation, and an interface a junior engineer can read in five minutes.
- Agentic systems — planning loops, tool registration, memory, evaluation harnesses
- Retrieval pipelines — ingestion, chunking, vector search, retrieval evaluation that you can actually measure
- Product platforms — SaaS metrics, feature gating, configuration, dashboards
- Developer tools — tiny zero-dependency utilities that I needed and couldn't find done well
| Layer | Repositories | What's inside |
|---|---|---|
| Agentic AI | agent-framework · rag-pipeline · prompt-eval |
ReAct planning, tool registration, episodic memory; production RAG with pluggable chunking and retrieval evaluation; structured prompt evaluation harness |
| AI tooling | ai-toolkit · token-counter · schema-gen |
Embeddings comparison, model output diffing, prompt benchmarking, multi-provider tokenizer + cost estimation, Pydantic v2 model generation from JSON samples |
| Product infrastructure | startup-dashboard · feature-flags · config-loader · task-queue · rate-limiter |
SaaS metrics dashboard (MRR/churn/LTV/CAC), TypeScript feature-flag SDK, type-safe config loader, in-memory priority queue, token bucket + sliding window rate limiter |
| Developer utilities | pathmask · globmatch · crontab-lite · semverlite · stringcase · markdownlint · json-differ · hexdump |
Glob path matching, fnmatch++ with extglob, cron parsing/evaluation, semver, case conversion, markdown linting, JSON diff, canonical hexdump — all zero-dependency, all 100% test coverage |
I'm building a portfolio of small, polished, zero-dependency libraries — each spec'd, test-driven, self-reviewed, and shipped with a complete README. Current count: 34+ utilities spanning data structures, parsers, formatters, and CLI helpers, with target line coverage at 100%.
A few that stand out:
bitvec— packed bit-vector with set ops and slicingcrontab-lite— cron expression evaluator withnext_fire/prev_firerangeset— non-overlapping integer-range set algebratrie— prefix tree with autocompletelevendist— Levenshtein, Damerau-Levenshtein, Jaro, Jaro-Winkler, LCS in one placeglobmatch— fnmatch superset with extglob, POSIX classes, and**globstartsparser·base62-ts— tokenizer/parser and Base62 codec for TypeScript
AI/ML Python · RAG · embeddings · vector search · prompt evaluation · agent workflows
Backend FastAPI · Node.js · Express · PostgreSQL · Redis · WebSockets
Frontend Next.js · React · TypeScript · Tailwind CSS · Recharts
Infrastructure Docker · GitHub Actions · Vercel · Linux · shell automation
Product SaaS metrics · feature flags · developer experience · enterprise AI adoption
Quality TDD · 100% line-coverage default · type-safe APIs · zero-dependency by preference
I prefer software that proves itself in use: clear interfaces, direct documentation, small testable modules, and graceful failure paths. My goal is not just to make AI demos that look impressive on stage — it is to turn intelligence into dependable product behavior.
Every repository above follows the same five-axis review: correctness · security · readability · performance · maintainability. Functions are short, tests are first-class, errors are typed, and dependencies are minimal. If you can't read it end-to-end in a single sitting, it's too big.
Open to collaborations on AI tooling, agent systems, product infrastructure, and enterprise AI transformation.
If you find any of these useful, ⭐ the repo — it helps me decide where to invest next.
AI-driven digital transformation leader | Enterprise AI product builder | Agentic systems architect
I build AI-native products that move from idea to usable system: agent workflows, RAG pipelines, developer tools, SaaS dashboards, and automation infrastructure.
I work at the intersection of product strategy, AI engineering, and enterprise transformation. My public GitHub is a lab for compact, practical systems: tools with clear APIs, readable docs, and implementation choices that can survive outside a notebook.
- Building agentic systems with planning loops, memory, tools, and evaluation
- Designing retrieval pipelines that can be measured, tuned, and shipped
- Turning product metrics, operations, and governance problems into usable software
- Keeping projects small enough to understand and strong enough to extend
| Area | Repositories | Why They Matter |
|---|---|---|
| Agentic AI | agent-framework, rag-pipeline, prompt-eval | Core building blocks for agents, retrieval, prompt evaluation, and production-grade AI workflows |
| AI tooling | ai-toolkit, token-counter, schema-gen | Practical CLI and library tools for prompt work, token budgeting, model output comparison, and structured data generation |
| Product systems | startup-dashboard, feature-flags, config-loader | SaaS analytics, runtime configuration, and feature-control foundations for product teams |
| Infrastructure utilities | task-queue, rate-limiter, dep-audit | Queueing, throughput control, and dependency risk checks for resilient services |
| Developer tools | markdownlint, hexdump, crontab-lite, json-differ | Small, well-scoped utilities for everyday engineering work |
AI/ML Python | RAG | embeddings | vector search | prompt evaluation | agent workflows
Backend FastAPI | Node.js | Express | PostgreSQL | Redis | WebSockets
Frontend Next.js | React | TypeScript | Tailwind CSS | Recharts
Infrastructure Docker | GitHub Actions | Vercel | Linux | shell automation
Product SaaS metrics | feature flags | developer experience | enterprise AI adoption
I prefer software that proves itself in use: clear interfaces, direct documentation, small testable modules, and graceful failure paths. The goal is not just to make AI demos impressive; it is to turn intelligence into dependable product behavior.
Open to collaborations on AI tooling, agent systems, product infrastructure, and enterprise AI transformation.