KENGUTECH: enterprise Java systems and applied LLMs

Enterprise systems for millions of requests, designed by a researcher.

LEAD ENGINEER · PHD CANDIDATE · AI ENGINEER

KenguTech plans and ships enterprise-grade Java microservice systems built to handle millions of requests, alongside applied LLM solutions, for businesses that need an engineer who has shipped both.

See capabilitiesRead research
§01/ architecture

Beyond writing services.

Lead engineer and software architect, with a decade of shipping systems that handle millions of requests under real production load.

KenguTech is not an outsource shop that ships features. We design the systems that other teams build on: domain boundaries that hold, resilience patterns that survive bad days, and high-availability topologies sized for actual load.

  • 01 / Lead engineer

    Technical leadership.

    A decade leading multi-team engagements. Architecture decisions that hold up under review, code that survives the next ten people who touch it, and the kind of mentoring that scales the team faster than headcount.

  • 02 / Software architect

    System design.

    Domain-driven boundaries, event-driven integration, capacity planning, observability. We design the systems other teams ship, sized to the traffic they actually face rather than the traffic the slide deck claimed.

  • 03 / Built for scale

    Millions of requests.

    Production patterns from a decade on-call: multi-region active-active topologies, circuit breakers and bulkheads, idempotency guarantees, SLO definition with real error budgets. Five-nines availability is a target, not an aspiration.

§02/ capabilities

What we build.

Five service tracks and one research lab. Nothing aspirational here: only work we ship in production or publish in peer review.

§03/ affiliations

Read by peers.

A studio's credibility is who works with it. Ours sits at the intersection of industry and academia.

Obuda UniversityDoctoral School of Applied Informatics & Applied Mathematics

PhD candidate (2024–2028), Assistant Lecturer at the John von Neumann Faculty of Informatics

HUN-REN AI1ScienceHUN-REN × Obuda University consortium

Strategic AI-research partnership building agentic-discovery tools that support the full scientific lifecycle: hypothesis generation, research planning, data analysis, and publishing.

§04 / currently writingFrom the desk.01 / 06
in progress2026· In preparation

benchgap

Mind the Gap: Explicit Cross-Benchmark Calibrations for Estimating Missing LLM Evaluation Scores

Most LLMs are evaluated on only a few benchmarks: across 401 models and 146 benchmark versions, coverage is just 8.3%. benchgap estimates the missing scores without running any evaluations, calibrating each benchmark against same-capability benchmarks with monotone curves selected by leave-one-out cross-validation, and attaching an error and confidence level to every estimate. In a 15% holdout it fills 71% of masked cells at 4.6 pp MAE, 23% lower than a matrix-completion baseline. It also quantifies the harness tax, the disagreement between evaluation setups on the same models. Estimates are updated daily at benchgap.net.

Try the toolSourceAll works
§05 / talk to us

Need an engineer
who has shipped
both enterprise systems
and peer-reviewed research?

T&M engagements, dedicated teams, or applied LLM consulting. We answer within 48 hours.

Open contact form