Looking for a job

Aleksey Tkachenko

@alekseytkachenko

AI engineer at Ocado who built a Model Gateway giving eight product teams real-time access to in-house models.

At Ocado, I built an internal Model Gateway with routing, authentication, and output-quality logging, giving eight product teams real-time access to in-house models without sensitive data leaving the environment. I also delivered a platform of customer-support assistants and fine-tuned production LLM agents, resolving half of a previously failing task category.

At CODE AI, I built a Python coding-agent MVP using graph-based RAG and AST trees, improving Pass@10 on RepoEval from 20% to 64%. Earlier, at InData Labs, I improved critical-defect recall from 0.72 to 0.94 and optimized model inference to run at 120 parts per minute.

Experience

Work history, roles, and key accomplishments

Senior AI Engineer

Ocado

Apr 2024 - Jul 2026 (2 years 3 months)

Built an internal Model Gateway on vLLM stack, delivering real-time access to in-house models for 8 product teams. Delivered a platform of 30+ customer-support assistants and fine-tuned production LLM agents, resolving half of previously failing task categories.

Python fastAPI Kafka VLLM LangGraph LoRAs SFT Reinforcement Learning MLFlow Airflow

NLP Engineer

CODE AI

Sep 2022 - Apr 2024 (1 year 7 months)

Built an MVP machine learning coding agent using graph-based RAG with AST trees, improving Pass@10 on RepoEval from 20% to 64%. Implemented a caching mechanism that reduced graph updates from 15 minutes to 30 seconds.

Python RAG AST Graph RAG RepoEval Caching

Machine Learning Engineer

InData Labs

Feb 2020 - Aug 2022 (2 years 6 months)

Raised recall on critical defects from 0.72 to 0.94 at a 6% false-positive rate. Optimised detection models by exporting to ONNX, enabling pipeline operation at 120 parts/min instead of 100 parts/min.

Machine Learning Onnx PyTorch AWS Sagemaker Airflow MLFlow PSI Monitoring