JANUSZ LENKIEWICZ
AI ENGINEER · PSYCHOLOGIST · 100% REMOTE

I build AI systems where the human has the final say

SEE PROJECTS →
14 YEARSof commercial software engineering
2 × MAdevelopmental psychology · pedagogy (Adam Mickiewicz University)
7production AI systems of my own — design, code, deploy, maintain
200K+lines of enterprise TypeScript across the codebases I work in daily — plus ~35K lines of my own production Python

The model is just a tool. The edge is understanding people.

I'm an LLM application engineer: I don't train models from scratch — I integrate them with real systems: backend, database, authorization, deployment, cost and monitoring. Behind me are fourteen years of commercial software engineering, from e-learning and banking-sector systems to enterprise front-end in international projects for aviation, e-commerce and pharma.

The second pillar of my profile is unusual: two master's degrees — in developmental psychology and pedagogy — and five years of university lecturing. That's why I specialize in human-in-the-loop systems: ones where AI supports a person's decision instead of making it for them. I know why people distrust systems built to "help" them — so I design AI that supports a human decision instead of quietly taking it over.

I built every system below myself and use them daily. This is deliberate dogfooding — I carry the cost of design mistakes on my own skin, before anyone else does.

EXPERIENCE
Since 2025independent AI engineer — 7 production LLM systems of my own: multi-agent, RAG, realtime voice, evaluation
2017 – nowSii Poland, senior front-end developer — enterprise projects for a US airline, international e-commerce and pharma; hundreds of thousands of lines of TypeScript across these codebases; code review, CI/CD, direct work with business and UX teams
2016–2021university lecturer, SGGW Warsaw — original postgraduate course, from syllabus to lectures
2016–2017ineoSMART, full stack — CRM for the banking sector, e-learning platforms for mBank and the National Bank of Poland
2011–2016mediaKURSY, full stack — course management systems, Moodle modules (incl. cooperation with Adobe), decision games for NGOs
Education2 × MA, Adam Mickiewicz University (developmental psychology · pedagogy) · CS50 AI with Python, Harvard (2023)
AI stackClaude API + Agent SDK · Gemini (incl. Live) · MCP · structured outputs · RAG · evals
EngineeringPython (FastAPI) · TypeScript/Node · React · Swift · Postgres · GCP Cloud Run · Docker
PROJECTS — BUILT, DEPLOYED, MAINTAINED
01

Trader AI Assistant

Mentoring platform · multi-agent

A production platform where an AI mentor guides the user according to their own written methodology — instead of generic advice. Multiple cooperating agents on the Claude Agent SDK, prompts assembled dynamically from the methodology and user history, integration with a brokerage account (auto-journaling, a live risk-rule guardian). The system runs on real money — AI mistakes carry a real cost here.

Claude Agent SDK · Node/Express · React + TS · Postgres · Firebase · Cloud Run
MORE — ARCHITECTURE & COMPETENCIES
  • End-to-end architecture. React 19 + TypeScript SPA (Vite, MUI, Zustand), Firebase Auth/Hosting/Storage, Express + TypeScript containerized on Google Cloud Run, PostgreSQL on Supabase. All domain data flows through the backend API; model API keys live server-side only.
  • LLM layer with two providers (Anthropic Claude and Google Gemini) and backend-level dispatch. Every prompt is assembled from blocks: trader profile, session history, market context and the user's own written methodology — setups, checklists and risk rules are data in the system, so the mentor enforces the user's rules, not generic advice.
  • Multi-agent layer on the Claude Agent SDK running as background jobs, with an isolation rule I designed: an analyzing agent cannot see the mentor's earlier assessments before issuing its own.
  • Model ops. A central model registry in the database (engines swappable without a deploy), request limits per subscription plan, BYO API keys, and a cost-driven default-model choice — comparable quality at a fraction of the price.
  • Production discipline. 100+ numbered SQL migrations, Vitest unit tests, Playwright E2E on ephemeral accounts, and procedures forged on real incidents — e.g. an env-replacing deploy flag led to a merge-and-verify procedure; a broker API returning HTTP success on rejected operations taught me to validate response bodies, not status codes.
COMPETENCIES: multi-agent orchestration · personalized RAG · prompt engineering · LLM ops & cost control · testing (unit / E2E) · GCP serverless
02

Deep Mentor

Personalized retrieval (RAG)

The mentor's long-term memory layer: the model reads the user's personal knowledge base — notes, lessons, hypotheses — and cites it in its analyses, instead of answering from general knowledge. Base seeding, digest in prompts, hard control over inference cost.

Retrieval over a personal knowledge base · context condensation · cost engineering
MORE — ARCHITECTURE & COMPETENCIES
  • Per-user wiki memory in the database: agent pipelines analyze the user's day and write notes into their personal knowledge base, which the mentor then cites in its analyses.
  • Retrieval designed for cost. Base seeding, digests in prompts instead of raw dumps, expensive deep runs only on demand — inference cost is a first-class design constraint, not an afterthought.
  • Two feed modes. The knowledge base can be filled by the platform's own automations or by an external pipeline compiling and pushing notes from the user's local Obsidian vault.
COMPETENCIES: RAG / context engineering · background agent pipelines · inference cost engineering · knowledge-base design
03

audit-mentor

LLM quality evaluation

A harness for the cyclical audit of the AI mentor's answers: an error taxonomy in four classes, deduplication of findings by root cause, a remediation backlog with locations in the code. Organizing principle: deterministic code computes the metrics, the model only interprets — an LLM doesn't grade itself on data it could distort.

Error taxonomy · human review loop · evaluation harness
MORE — ARCHITECTURE & COMPETENCIES
  • Cyclical audit (~every two weeks) of the mentor's production answers. Every finding carries an evidence quote and a root-cause hypothesis — no vague "could be better" notes.
  • Regression backlog. Findings are deduplicated by root-cause signature, an occurrence counter works as the priority weight, and each entry points to path:line locations in the codebase.
  • Division of labor by design. Deterministic code computes the metrics, the model interprets them, and the human reviews findings after the fact and kills false positives.
COMPETENCIES: LLM evaluation · error taxonomy design · human review loops · observability & regression tracking
04

Prompter & Murmur

Realtime voice AI · macOS

Native apps for live conversations: streaming transcription of both sides of the conversation, simultaneous translation, and contextual prompts from a personal knowledge base. The main engineering challenge: the latency budget of the audio → text → LLM → response pipeline, across two parallel model sessions.

Swift · Apple SpeechAnalyzer · Gemini Live · low-latency engineering
MORE — ARCHITECTURE & COMPETENCIES
  • Streaming transcription of both sides of a conversation with two model sessions running in parallel; the core engineering work is the latency budget of the audio → text → LLM → response pipeline.
  • Hybrid speech stack. On-device Apple SpeechAnalyzer where the language is supported, Gemini Live where it isn't — chosen per language after real spikes on the target OS, not from documentation alone.
  • Contextual prompt cards retrieved live from a personal knowledge base during the conversation; simultaneous translation for cross-language calls.
COMPETENCIES: realtime voice AI · low-latency pipeline design · Swift / native macOS · Gemini Live · streaming STT
05

Scout

Research agent · MCP

A market-analysis agent with a hard autonomy boundary: it analyzes, suggests and alerts, but the decision always stays with the human — zero auto-trading, no exceptions. Deterministic metrics plus a narrative layer from web research, a walk-forward backtesting lab, tool integrations over the Model Context Protocol, including a bridge to TradingView Desktop.

MCP · walk-forward backtesting · exchange REST APIs · Python
MORE — ARCHITECTURE & COMPETENCIES
  • A market radar that separates roles: deterministic code computes volatility metrics, a narrative layer from web research adds context — the model never invents the numbers it reasons about.
  • "Blind" analyses. The agent produces its analysis without seeing the human's view, and only then are the two diffed — a learning loop deliberately designed to avoid anchoring.
  • Setup Lab: walk-forward backtesting of strategies with an append-only experiment journal; tool integrations over MCP (including a bridge controlling TradingView Desktop) plus direct exchange REST APIs.
COMPETENCIES: MCP tool integrations · agent design with autonomy boundaries · walk-forward backtesting · Python data stack
06

Python trading system

~35K lines · case study

A production asynchronous service: a layered architecture with the execution service as the sole owner of orders, centralized risk configuration, unit tests on the money path, demo mode as the safe default. Alongside it — a signal engine with LLM-based signal scoring in a live loop. Private code, discussed as a case study in interviews.

FastAPI · Pydantic · asyncpg · websockets · SQLAlchemy · Postgres
MORE — ARCHITECTURE & COMPETENCIES
  • ~30.7K lines: an async FastAPI service with a layered architecture — connectors → analysis → an execution service as the sole owner of orders. Centralized risk configuration, demo mode as the safe default.
  • Tests where money moves. Unit tests concentrated on the money path: loss limits, pause-after-loss, signal-age validation.
  • ~5.1K lines alongside: a 12-rule volume-price signal engine with a live candle-by-candle loop and LLM-based signal scoring; SQLAlchemy/Postgres persistence.
COMPETENCIES: Python service architecture · async / websockets · risk-logic testing · LLM signal scoring · SQLAlchemy / Postgres
07

Atelier

AI in professional development · EdTech

A year-long professional-development program for someone entering the styling profession — led by an AI agent under my subject-matter supervision. A daily newsletter generated from web research and the participant's personal profile, audio lesson series released one per day, a dashboard with the day's task, and a feedback loop: the participant's entries feed back into the generator of the next issues. The psychology of adult learning translated directly into system architecture.

Headless agent (cron) · feedback loop · TTS · curriculum
MORE — ARCHITECTURE & COMPETENCIES
  • A headless agent on a daily cron generates each issue from web research, the participant's profile and the program's weekly themes; an anti-repetition rule makes it read the last 14 issues before writing a new one.
  • Audio lessons and a dashboard. TTS lesson series released one per day from a queue; a dashboard with the day's task; the participant's entries and check-ins feed back into the next issues.
  • Curriculum built on psychology. A year-long program map designed around the psychology of entering a new profession — dosing, celebrating streaks, support instead of accounting.
COMPETENCIES: agentic automation (headless / cron) · curriculum design · TTS pipeline · feedback-loop design
COMPETENCIES — LLM / AI ENGINEERING
IN PRODUCTION — WITH PROOF
LLM application developmentProduction applications on Claude API + Agent SDK and Gemini (incl. Live) since 2025 — backend, auth, deployment, cost, monitoring.Proof: Trader AI Assistant · Prompter & Murmur
Multi-agent orchestrationCooperating agents as background jobs: handoffs, structured outputs, tool-calling, isolation rules between agents.Proof: Trader AI Assistant
Personalized retrieval (RAG)Retrieval over the user's own knowledge base: seeding, digests in prompts, hard inference-cost control.Proof: Deep Mentor
LLM evaluation & observabilityError taxonomy, evidence-backed findings, root-cause deduplication, human review loops, regression backlog — one of the rarest skills on the market.Proof: audit-mentor
MCP — Model Context ProtocolTool integrations over MCP, including a bridge controlling TradingView Desktop; already a must-have in job listings.Proof: Scout
Prompt engineeringProduction prompts assembled dynamically from blocks: user profile, methodology, session history, market context.Proof: Trader AI Assistant
Realtime voice AIStreaming transcription, simultaneous translation, latency budgets across parallel model sessions.Proof: Prompter & Murmur
Production PythonAsync FastAPI services, websockets, layered architecture, unit-tested risk logic on a live exchange account.Proof: trading system, ~35K lines
Cloud & DevOpsGCP Cloud Run (continuous revisions), Firebase, Supabase/Postgres, Docker, CI/CD — plus 100+ SQL migrations in production.Proof: every project above
Human-in-the-loop designAutonomy boundaries, review loops, AI adoption and enablement — backed by two master's degrees and five years of lecturing.Proof: all of the above + academic record
IN PROGRESS — LEARNING BY BUILDING

The market sometimes names my patterns after libraries I haven't used yet. I say that openly — and I close each gap inside a live project, not with a course certificate.

LangGraphThe agent patterns I already run on the Claude Agent SDK, re-implemented in the market's standard library — as an analysis agent in my trading bot.
Vector retrieval (pgvector)Adding an embeddings leg to Deep Mentor's retrieval and naming the existing pipeline in market terms: chunking, embeddings, reranking.
Langfuse & promptfooIndustry tooling layered onto the evaluation harness I already run in production.
MLflowExperiment tracking for the walk-forward backtesting lab — the experiment discipline already exists, the tool is being added.
Classical ML (scikit-learn)A volatility-regime classifier trained on my own market data — on top of CS50 AI (Harvard) foundations.
Spoken English B2+Daily practice with my own realtime voice tools — reading and writing are fluent, speech is in active training.
WHERE PSYCHOLOGY MEETS ENGINEERING
SYCOPHANCY

I detected sycophancy in my own AI mentor: it anchored on the user's emotional self-assessment and confirmed it, instead of confronting it with data. As a psychologist I know the difference between supporting and agreeing — which is why a mentor that doesn't just agree has to be designed.

AUTONOMY BOUNDARIES

The most important architectural decisions are about what the system must not do. Zero auto-trading in the market agent. The only mutation on the broker's side is a protective daily lockout. Designing such boundaries requires understanding a person under pressure — their fatigue and their impulses.

HUMAN–MODEL DIVISION OF LABOR

Deterministic code computes the metrics, the model interprets, the human reviews after the fact. A deliberate human-review-loop architecture — instead of trusting the model to grade itself.

COLLABORATION

I'm looking for projects at the intersection of AI engineering and people: mentoring and educational products, human-in-the-loop systems, LLM applications from architecture to production.

B2B contract, fully remote. I reply to every concrete message.