Kostiantyn

KOSTIANTYN BAKULIN
AI Engineer & Technical Lead | LLM & Agent Systems
[відкрити контакти](див. вище в блоці «контактна інформація») | github.com/KostiantynBk | Spain / Remote

AI engineer building Python backends, multi-agent systems, voice agents, RAG pipelines, and business automation.
Commercially shipped trading-platform services and led architecture for a multi-agent product — specs, interface
contracts, and review for a distributed team. Same patterns apply outside trading: ops, support, and customer-facing
workflows.

EXPERIENCE

Technical Lead / Project Manager - AI Agent Systems May 2026 - Sep 2026
Mediaverse Pte. Ltd. | Independent Contractor (Promoted from AI Systems Developer) | Remote
Owned architecture for a multi-agent trading platform: agent roles, tool boundaries, backend services, and the
path from research signals into order/journal workflows.
Turned trading-domain requirements into specs and interface contracts; set implementation order and reviewed
Python changes for a distributed team.
Ran delivery in Monday.com with MCP so coding agents could create, update, and close tasks from the repo
instead of a stale board.
Introduced a Hermes agent with a memory layer so new joiners only retrieved project context they were allowed
to see (NDA-gated).

AI Systems Developer Mar 2026 - May 2026
Mediaverse Pte. Ltd. | Independent Contractor | Remote
Built Python services and agent workflows for a multi-agent trading platform, wiring LLM tools to
market/application data and automation steps.
Translated desk requirements into backend components and specs for ingest, research, and agent hand-offs.
Implemented statistical and ML utilities with NumPy, SciPy, statsmodels, and scikit-learn for residualization,
regressions, and time-series checks used by agent tools.
Worked spec-to-PR with Codex, Claude Code, and Cursor, using Perplexity for research against written contracts.

Backend Engineer Sep 2024 - Feb 2026
Voxum | Independent Contractor | Hybrid
Built FastAPI services for strategy automation: live market ingest, signal workflows, and order/journal persistence.
Ingested market data over WebSocket and FIX with asyncio; published events on Redis Streams and cached hot
state in Redis.
Stored ticks/bars in TimescaleDB and orders in Postgres; kept research datasets as Parquet on S3 and queried
them with DuckDB/Polars.
Packaged services with Docker, deployed on Kubernetes, and added Prometheus/Grafana dashboards and alerts
for ingest lag and service health.

KOSTIANTYN BAKULIN / AI ENGINEER & TECHNICAL LEAD 1

SELECTED PROJECTS

Multi-Agent Research Desk
n8n / OpenAI API / AI Agent Nodes / Sub-Workflows / HTTP / JavaScript
Route research requests to specialized web-research, calculation, and summarization agents.
Merge sourced results, handle conflicting findings, and use cost-aware model escalation.

DocsRAG - Documentation Assistant GitHub ↗
Python / OpenAI API / Chroma / Sentence-Transformers / LangChain / FastAPI / SQLite
Retrieve relevant documentation passages and produce schema-validated answers with citations.
Compare chunking and reranking strategies using retrieval metrics and LLM-based evaluation.

Terminal AI Trading Research Agent GitHub ↗
Python / OpenAI Responses API / Pydantic / SQLite / Binance API / Ubuntu
Analyze crypto market data and generate structured long, short, watch, or no-trade research decisions.
Store market history and reports in SQLite while supporting backtesting and systemd deployment.

Observable AI Content Pipeline
n8n / OpenAI API / Langfuse / Structured Output Parser / Google Sheets
Generate structured business content while tracing model inputs, outputs, latency, tokens, and cost.
Compare prompt variants and run scheduled evaluations to detect quality regressions.

TECHNICAL SKILLS

Backend & data: Python, FastAPI, Pydantic, asyncio, Redis, PostgreSQL, TimescaleDB, SQLite, DuckDB, Polars
Trading & research: WebSocket, FIX, NumPy, SciPy, statsmodels, scikit-learn, Parquet, S3
Agents & RAG: OpenAI Responses API, n8n, MCP, Vapi.ai, LangChain, Chroma, Sentence-Transformers
Delivery: Docker, Kubernetes, Prometheus, Grafana, systemd, Ubuntu

EDUCATION

B.Sc. in Data Science (Frozen) 2024 - Present
Eindhoven University of Technology (TU/e) | Netherlands
Relevant coursework: Advanced Machine Learning, Statistics, Data Structures, Fundamentals of Data Analytics,
Linear Algebra, and Discrete Structures.

CERTIFICATIONS

Kaggle: Intro to Machine Learning, Intermediate Machine Learning, Feature Engineering, Intro to Deep Learning,
Computer Vision.
University of Michigan / Coursera: Introduction to Data Science in Python; Applied Plotting, Charting & Data
Representation in Python.

LANGUAGES

English: C1 | Ukrainian: Native | Russian: Native

KOSTIANTYN BAKULIN / AI ENGINEER & TECHNICAL LEAD 2