Szymon Klempert

Software Engineer

Software engineer with 4+ years of experience in regulated digital health: Python backends, EHR integrations and AWS infrastructure. Interested in distributed systems, LLMs and other emerging technologies. Big fan of well-organized data.

Experience

HTD Health

Jul 2022 – Present

Software Engineer

Behavioral Health Intervention Platform

2025 – Present

Ruby on Rails, React, Epic, FHIR, HL7 v2

  • Lead full-stack developer, and earlier the sole developer, of a 6.5 year old HIPAA-regulated platform used in clinical trials. Own the product side too: client communication, roadmap and feature proposals.
  • Adapted the platform to run a dedicated instance per health system customer. Created a repeatable onboarding process that covers each site’s EHR integration.
  • Brought the platform up to current standards: moved it from a legacy EC2 setup to containers on AWS ECS.
  • Initiated and led an AI-powered security review of PHI exposure across the platform.
  • Enhanced a pretty dead E2E suite with an AI-enabled pipeline that drafts new scenarios from merged PRs.

Selected client projects

2022 – 2025
  • Designed the event-driven layer that keeps two HIPAA-regulated portals in sync with Salesforce.Python, FastAPI, AWS
  • Built an identity resolution pipeline that links over 1 billion clinician records.Python, Snowflake, Splink
  • Ran the technical discovery for an AI startup’s EHR integration, with proofs of concept against Epic and Cerner.Python, SMART on FHIR, Redox

Education

MSc, Computer Science and Intelligent Systems

Feb 2024 – Oct 2026

AGH University of Krakow

Thesis: Comparison of Monolithic and Microservice Architecture in the Design and Implementation of Applications for Processing Continuously Updated Data Volumes

  • Built a monolith and a five-service RabbitMQ version of my own price-tracking pipeline and benchmarked them under an equal CPU and memory budget.
  • At the same throughput, the microservices used ~1.8× the CPU and ~2.4× the memory; scaling only the measured bottleneck raised throughput ~2.5×.

BEng, Computer Science and Intelligent Systems

2020 – 2024

AGH University of Krakow

Thesis: System Using Elasticsearch Software to Store and Index Data Scraped from the Internet

  • Web scraping pipeline (Python, Scrapy) that collects real-estate listings from Otodom, normalizes them (typed fields, geocoding) and bulk-indexes them into Elasticsearch.
  • Served through a FastAPI search API (attribute, fuzzy and geospatial search), with a React app and a Kibana dashboard for visualization.

Projects

Turns the chaos of second-hand listings scattered across marketplaces and groups (looking at you, Facebook) into structured, scored, comparable data.

  • Takes raw screenshots of Facebook posts (listings) through a Telegram bot.
  • Returns an analysis of the listing based on predefined criteria and scores it. Drafts a WhatsApp message that asks the seller for missing details, and uploads everything to a Google Sheet.
  • The LLM writes the analysis, but the code computes the score, so the same listing always gets the same score.
  • Campervans are the only domain for now (as I’m looking for one, DM me if you’re selling ;) Any other domain needs just a few files:
    • knowledge base: scoring instructions and known faults
    • personal criteria: what I’m looking for

Python, Telegram Bot API, LLM APIs, MCP, Google Sheets

Skills

Achievements