Tom Artiom Fiodorov
Senior Backend & AI Engineer · 10+ yrs (5+ in crypto)
Senior AI / backend engineer building production LLM, RAG, and agentic systems.
Las Palmas, Spain — remote or relocation
+34 682 039 872
tom@fiodorov.es
github.com/afiodorov
EU citizen (Lithuanian)
Things I've Built
- hn.fiodorov.es — production RAG over 12M+ Hacker News comments: LangGraph + DeepSeek answer pipeline; custom Rust vector service (axum/rayon, mmap) replacing pgvector; ONNX embeddings, Redis caching, FastAPI SSE streaming (source).
- lyrics.fiodorov.es — LangGraph workflow (LangChain, Tavily, Redis memory) that finds, formats, and translates song lyrics.
- imdb-sql.fiodorov.es — the full IMDB dataset queryable with SQL in the browser via React + WebAssembly; no backend at all.
- livetranslate.fiodorov.es — real-time speech captioning & translation in the browser (React, TypeScript).
Experience
Kraken (Remote)
Jan 2021 – Mar 2026
AI Engineer (Nov 2025 – Mar 2026)
- Built and shipped a company-wide LLM personalization system delivering tailored recommendations and performance insights to 100k+ users at launch — surfacing each user's PnL and trading activity in natural language (e.g. "your portfolio outperformed BTC holdings last month").
- Designed the feature-engineering pipeline extracting and transforming signals from the data lake into grounded, structured prompts — the layer that kept outputs accurate and personal rather than generic.
- Owned prompt design and built an LLM-as-a-judge evaluation harness (on Gemini Flash) scoring outputs for conciseness, factuality, and financial-advice safety — each dimension calibrated against ~50 manually-labeled examples — enabling confident iteration at scale.
- Served the system through gRPC microservices in production.
Data Engineer (Mar 2023 – Nov 2025)
- Built the pipeline decrypting withdrawal/deposit data inside restricted zones (Spark, Airflow, EMR) — now a heavily-used compliance dataset.
- Cut AWS spend by $10k+/month with in-place Apache Iceberg conversions and Athena query optimisation.
- Flattened deeply nested semi-structured data with recursive-SQL pipelines, materially improving internal dataset usability.
Go Backend Developer, CryptoWatch (Jan 2021 – Mar 2023)
- Built high-throughput, real-time streaming microservices in Go (gRPC, Protocol Buffers) for live trading data.
- Ran services on Kubernetes; optimised data-processing logic against AWS Aurora (PostgreSQL).
Glassnode (Remote) — Blockchain Backend Developer
Jul 2019 – Nov 2020
- Engineered 60+ novel blockchain econometrics in Go and advanced SQL — core components of Glassnode's market-intelligence platform.
- Designed an augmented red-black tree (open-sourced) that processes the full historical Bitcoin ledger daily, in minutes.
RokkerX — Data Scientist (part-time)
2024
- Built real-time AI services for live speech captioning & translation (Python, FastAPI).
Datapane — ML Engineer
2018 – 2019
- Shipped time-series forecasting, churn, and anomaly-detection models to production.
Ravelin — Junior Data Scientist
2016 – 2018
- Automated GCP MLOps pipelines behind real-time fraud-detection ensembles.
Education
UCL — MPhil in Probability (2012–2016), published in Electronic Journal of Probability; BSc in Mathematics, First Class Honours (2008–2011).
University of Cambridge — Master of Mathematics, MMath (2011–2012).