Technical Lead, Evaluations Platform at InPost

Seniority in posting: senior

Language

Setup: Hybrid
Location: Krakow, Lesser Poland Voivodeship
Type: Contract
Level: lead/mgmt
Posted: Sep 17, 2026

Description

InPost is hiring a technical lead for its Von Halsky Evaluations Platform, which evaluates and gates conversational AI releases. The role owns the LLM-judge pipeline, eval datasets, root-cause analysis, anomaly detection, eval-driven development, and the interface between engineering and product and business stakeholders. It is a hands-on technical leadership position without line management, with hybrid work in Warsaw or Kraków and a B2B-type contract. Candidates need at least five years of production software experience, two years on LLM-based systems, strong Python and cloud engineering fundamentals, and experience evaluating generative systems.

Required Skills

Python
Engineering fundamentals
Testing
CI/CD
Containers
Observability
Cloud
LLM evaluation
Prompt engineering
SQL
Data at scale
AI development tools

Additional Information

Required Previous Experiences: 5+ years building and running production software
2+ years on LLM-based systems used by real users
Experience evaluating generative systems, including LLM-as-judge, human-label calibration, inter-annotator agreement, regression suites, and offline-versus-online divergence
Preferred Qualifications: Harness and loop engineering
Auto-improving systems
Adversarial robustness, jailbreak testing, red-teaming, or abuse and fraud detection
E-commerce, search, or recommendation domain experience

About InPost

InPost

e-commerce enablement platformLuxembourg, LU

InPost is an out-of-home e-commerce enablement platform in Europe, providing delivery through an out-of-home network, to-door courier and fulfillment services to e-commerce merchants. It describes its solution as cost-effective and greener for merchants, while giving customers more convenience. InPost was founded in Poland in 1999 and operates in 9 European countries through more than 40,000 APMs and almost 33,000 PUDOs. Its 2025 performance highlights include 1,364.8 million parcels delivered and 62.0 thousand active Automated Parcel Machines.