Christopher G.
Senior Software Engineer
Experience
Apr 2024 - Present
1 year 9 monthsSenior Software Engineer
Allegro
- Re-platformed the checkout cluster to Spring Boot 3 (MVC) with Kafka + Outbox/Saga, halving hot-path p99 420 ms → 210 ms (-50%) at 6.5k rps; validated in Grafana + Datadog APM and k6 load runs.
- Introduced Redis + Caffeine cache-aside for availability/pricing, cutting DB reads -38% and saving ≈$14k/quarter in IOPS; proven via RDS Insights, Redis INFO, and cost reports.
- Delivered zero-downtime blue/green deploys (Argo Rollouts, PDBs, readiness gates) for pricing API, taking deployment failure rate 3.1% → 0.6% and MTTR 27 → 9 min; evidence in Argo + PagerDuty timelines.
- Shipped OpenTelemetry traces + logs/metrics correlation across 12 microservices; cross-service debug time dropped 45 min → 12 min; measured in Jaeger/Tempo and on-call reports.
- Migrated legacy catalog UI from AngularJS to Angular 17 micro-frontends and built a React promo-rules editor (module federation, RxJS query, React Hook Form); LCP p75 3.2 s → 1.9 s, CLS 0.14 → 0.04; verified by Lighthouse + Web Vitals.
- Added feature-flagged A/B price rules (Unleash + Spring Boot starter) with Micrometer business counters; conversion on targeted SKUs +2.4%; experiment readout in internal experimentation platform.
- Tuned GC and connection pool (G1 → ZGC, Hikari), plus HTTP thread tuning; CPU -22%, container requests -18%, saving ≈$5.7k/month in infrastructure; shown in Kubernetes + Datadog dashboards.
Mar 2024 - Mar 2025
1 year 1 monthSenior Software Engineer
Artisan AI
- Built a low-latency inference gateway (Spring WebFlux + Netty, gRPC) in front of GPU workers; p95 ~180 ms at 2.1k rps, 99.9% availability; validated via Prometheus SLOs, Grafana, k6.
- Implemented RAG retrieval service (Java + pgvector + Redis Bloom) with re-rank; search F1 +13% and tail p99 520 ms → 260 ms; metrics from offline eval harness + APM traces.
- Orchestrated async embeddings/batch jobs (Kafka, Spring Cloud Stream, Outbox/Inbox), eliminating duplicate work and lowering reprocessing costs -37%; confirmed by Kafka consumer lag + cost reports.
- Added AI quality guardrails (prompt templates, input sanitization, semantic grounding, toxicity classifier) cutting hallucination rate -28% and policy violations -62%; measured by weekly eval set + human review.
- Containerized models on Triton + KServe with GPU autobatching + dynamic concurrency; throughput +41% per A100 and unit cost -24%; numbers from Prometheus GPU exporter + billing.
- Implemented security & compliance measures: mTLS, per-tenant quotas (Bucket4j + Redis), SBOM (Syft/Grype) + Cosign image signing; critical CVEs -92% in two quarters; trends from CI security gates.
- Built a React admin console and Vue-based prompt/playground with latency charts, eval diff views, and feature flags; PR cycle time 3.4 → 2.1 days (-38%) and post-release integration bugs -41%; GitHub/CI analytics.
- Optimized JVM footprint (JLink, CDS, tiered compilation) on Alpine base; image size -48%, cold-start -35% for serverless workers; validated in registry stats + cold-start traces.
Oct 2021 - Feb 2024
2 years 5 monthsDevOps Engineer
AioCare
- Rolled out a Kubernetes platform (three environments, eight node pools) with Terraform + Helm + Argo CD, cutting deploy time 45 min → 7 min (-84%) and enabling ~20 production releases/week; verified in Argo CD history and Jira releases.
- Built progressive delivery pipelines (blue/green + canary 5%→25%→100%) on GitHub Actions → Argo Rollouts, reducing failed deploys by ~72% and MTTR 38 min → 11 min; confirmed via PagerDuty + Grafana incident timelines.
- Introduced SLOs (p95 API < 250 ms, error budget 1%/30 d) with burn-rate alerts; lifted API uptime to 99.95% over 12 months and cut MTTD 12 min → 3 min; Prometheus/Grafana dashboards as evidence.
- Optimized cloud spend -28% YoY using HPA/VPA, spot nodes, storage lifecycle policies, and right-sizing; environment spin-up shrank from two days → two hours with reusable Terraform modules; validated by Cost Explorer/Billing + CI timestamps.
- Shipped end-to-end observability (OpenTelemetry traces, Prometheus metrics, Loki/ELK logs) and golden dashboards per service, lowering “unknown root cause” postmortems by ~60% and after-hours pages by ~50%; PagerDuty analytics + RCA tags.
- Hardened SDLC for medical data: SBOM + image signing (Syft/Grype, Cosign), OPA/Gatekeeper policies, Vault/KMS secrets; reduced critical CVEs in images by >90% within two quarters; security scan trends used for proof.
- Delivered DR runbooks and quarterly game-days achieving RPO ≤ 5 min and RTO ≤ 30 min via cross-region backups and infrastructure-as-code restores; results documented in DR drill reports.
Jul 2016 - Sep 2020
4 years 3 monthsSoftware Engineer
Spyrosoft
- Designed, developed, and maintained enterprise and consumer-facing applications for multiple international clients, including a fintech dashboard for BNP Paribas, an e-commerce platform for a retail client, and a healthcare management tool for a telemedicine startup.
- Delivered scalable solutions using React, Angular, and Vue on the front end, with FastAPI, Django, Java, and Go powering high-performance back ends, ensuring seamless integration, security, and excellent user experience.
Languages
English
AdvancedEducation
Oct 2014 - Mar 2016
Lublin University of Technology
Master’s Degree · Computer Science · Lublin, Poland
May 2010 - Sep 2014
Lublin University of Technology
Bachelor’s Degree · Computer Science · Lublin, Poland
Need a freelancer? Find your match in seconds.
Try FRATCH GPT More actions
Similar Freelancers
Discover other experts with similar qualifications and experience
- English
- Deutsch
2025 © FRATCH.IO GmbH. All rights reserved.