# Daniel Fridljand - Personal Website This page contains my professional biography, experience, and blog posts. --- # Daniel Fridljand **Software Consultant** at [TNG Technology Consulting](https://www.tngtech.com/en/) Software consultant at TNG focused on applied AI engineering, currently embedded as the sole AI engineer on a production agentic support-automation system (Claude, Temporal, MCP, DSPy) at an enterprise customer. Research background spanning ETH Zürich (computational oncology), Stanford School of Medicine (first co-author in Nature Medicine, 2024), and EMBL (statistical genomics). M.Sc. Mathematics with full marks; Yale exchange scholar. ## Interests - Applied AI - Machine Learning - Software Development ## Education - **M.Sc. in Mathematics**, 2023 — University of Heidelberg - **Exchange Scholar in Applied Mathematics**, 2023 — Yale University - **B.Sc. in Mathematics**, 2020 — University of Heidelberg ## Social Links - [Google Scholar](https://scholar.google.com/citations?user=SIoMbdMAAAAJ) - [GitHub](https://github.com/FridljDa) - [LinkedIn](https://www.linkedin.com/in/daniel-fridljand-8707a2208/) --- # Experience ## Software Consultant - Applied AI **TNG Technology Consulting** | Munich, Germany | Dec 2025 — Present * **Secure Desktop AI Agent (Jun 2026 – present)**: Core developer on a privacy-first desktop AI agent application combining a Tauri/React frontend with a Rust backend; full-stack contributions across a Bun/Rust monorepo covering the desktop UI, agent runtime, and secure auditable LLM tool execution. * **LLM-Powered Document Validation System (May–Jun 2026)**: Core developer in a 3-person team for a production-grade automated proposal review service, co-developing complex .docx document processing with automated comment insertion. * **AI-Powered Email to Order Parsing (Apr–May 2026)**: Designed a field-centric email-processing pipeline with FastAPI ingestion and Streamlit review dashboard; attachment-aware LLM processing with multimodal fallbacks for scanned PDFs, plus parallel candidate-resolution combining sender lookup, text extraction, and external address search. * **AI Customer Support Automation (Dec 2025 – Apr 2026, live in production)**: Sole AI engineer end-to-end on a cinema-ticketing SaaS — **770+ live B2C tickets** processed with **86.6% strict / 88.9% content-supported approval**; daily volume scaled 5x post-CEO showcase; 13+ feature improvements driven from live reviewer feedback. * Architected a hybrid deterministic + agentic workflow over 19 customer-intent categories using Temporal for durable orchestration, with human-in-the-loop approval via Signals and DSPy/GEPA prompt optimization. * Built a production evaluation suite (Langfuse: 6 evaluation types, 8 score metrics, 14-label outcome taxonomy) and a reproducible 22-stage Snakemake data pipeline with Microsoft Presidio PII detection for GDPR-compliant training data. * **Tech Stack**: Python, FastAPI, Streamlit, PydanticAI, Temporal, Langfuse, Snakemake, Docker, DSPy, MCP, Tauri, React, Rust. ## Software Consultant - Enterprise Modernization **TNG Technology Consulting** | Munich, Germany | Dec 2024 — Dec 2025 * Member of the platform team modernizing a mission-critical global supply-chain application (Java 8 → 17, JBoss → WildFly) in a multi-year transformation program. * Shipped a [JFrog Artifactory](https://jfrog.com/artifactory/) proxy in 3 days that reduced a recurring CI pipeline runtime from 8 hours to 30 seconds, saving developers ~1–2 hours per week each. * Established DevSecOps governance: integrated [OWASP Dependency-Check](https://owasp.org/www-project-dependency-check/) scans into CI, built Grafana dashboards for CVE monitoring, and migrated internal services from SOAP to REST with Keycloak. ## Research Data Analyst - Computational Oncology **ETH Zürich** | Basel, Switzerland | Feb 2024 — Sep 2024 * Developed novel Bayesian non-parametric methods (Hierarchical Dirichlet Process) for estimating mutational signatures in cancer genomes, extended to incorporate phylogenetic tree structures. * Analyzed single-cell whole-exome sequencing data from the Tumor Profiler Study (187 cells, 10 melanoma tumors), identifying eight latent mutational signatures. * Implemented hierarchical dependency structures in R such that signature distributions for child phylogenetic nodes are drawn from parent distributions, enforcing biological inheritance patterns. ## Research Data Analyst - Environmental Epidemiology **Stanford University School of Medicine** | Palo Alto, USA | Jul 2023 — Dec 2023 * **Nature Medicine Publication (2024)**: Led the statistical analysis as first co-author, quantifying air pollution's contribution to racial and socioeconomic mortality disparities in the US — [published in Nature Medicine](https://www.nature.com/articles/s41591-024-03117-0). * Engineered a big-data pipeline harmonizing 63+ million death records, satellite pollution estimates, and census demographics across 3,000+ US counties (1990–2016). * Implemented confounder-adjusted causal inference (DAGs, propensity scoring, multivariate regression), revealing >50% of the Black–White all-cause mortality difference is attributable to environmental factors. * Built and shipped an R Shiny analytical web application used directly by epidemiologists and policy researchers to explore 17-dimensional data and detect outliers. ## Exchange Scholar **Yale University** | New Haven, USA | Aug 2022 — May 2023 * Selected as one of two university-wide representatives for the year-long exchange program from the University of Heidelberg. * Grade: Honors (highest academic distinction at Yale); DAAD Stipend for academic excellence. * Coursework: Deep Learning, Geometric & Topological Methods in Machine Learning (Prof. Smita Krishnaswamy), Differentiable Manifolds, Statistical Methods in Human Genetics. ## Research Data Analyst - Statistical Genomics (Master's Thesis) **European Molecular Biology Laboratory** | Heidelberg, Germany | Oct 2021 — May 2022 * Developed **IHW-Forest**, a novel multiple-testing method using Random Forests for hypothesis weighting, increasing discovery power by >30% on a benchmark of 16 billion genetic association tests. * Optimized core splitting and weighting logic in C++ via Rcpp for high-performance processing of large-scale genomic data. * Presented at seven scientific events including Yale University and a competitively selected oral contribution at DAGStat 2022. ## M.Sc. in Mathematics **University of Heidelberg** | Heidelberg, Germany | Oct 2020 — May 2023 * Grade: 1.0 (summa cum laude, highest distinction). * Master's Thesis: "Better multiple Testing: Using multivariate co-data for hypothesis weighting", conducted at EMBL. * Awards: Gerhard C. Starck Foundation Stipend, Baden-Württemberg Stipend. ## Exchange Student **Hebrew University of Jerusalem** | Jerusalem, Israel | Sep 2019 — Mar 2020 * Graduate-level coursework in Functional Analysis, Algebraic Combinatorics, and Quantitative Models at the Einstein Institute of Mathematics. ## B.Sc. in Mathematics **University of Heidelberg** | Heidelberg, Germany | Oct 2017 — Sep 2020 * Grade: 1.4 (top 10% of cohort). * Bachelor's Thesis: "Online estimation of the geometric median in a Hilbert space". --- # Publications ## Disparities in air pollution attributable mortality in the US population by race/ethnicity and sociodemographic factors **Authors:** Pascal Geldsetzer, Daniel Fridljand, Mathew Kiang, Eran Bendavid, Sam Heft-Neal, Marshall Burke, Alexander H. Thieme, Tarik Benmarhnia **Journal:** Nature Medicine **Date:** 2024-07-01 **Publication:** [https://www.nature.com/articles/s41591-024-03117-0](https://www.nature.com/articles/s41591-024-03117-0) **Code:** [https://github.com/FridljDa/pm25_inequality](https://github.com/FridljDa/pm25_inequality) **Dataset:** [https://zenodo.org/records/10038691](https://zenodo.org/records/10038691) In the US between 2000 and 2011, over half of the gap in mortality between Black and non-Hispanic White adults can be explained by the fact that Black adults are, on average, more exposed and more susceptible to air pollution than non-Hispanic White adults. --- # Blog Posts ## How I Used Claude Design to Create a Hackathon Slide Deck A short post about using Claude Design to create a hackathon slide deck, with the full deck embedded below. **Published:** June 25, 2026 **Tags:** AI, Design, Hackathon **Link:** [View post](/post/claude-design-hackathon-slide-deck) ## Architecture of a First-Level Support Automation The four-step pipeline we landed on after iterating in production: a rooted DAG of intents, grouped extraction, weighted-scoring fetch, and a mostly-hardcoded planner with an explicit LLM fallback. **Published:** May 16, 2026 **Tags:** AI, Customer Support, Architecture **Link:** [View post](/post/first-level-support-automation) ## On the Future of Pull Requests in the Age of Agentic Coding **Published:** May 8, 2026 **Tags:** AI, Software Engineering, Agentic Coding **Link:** [View post](/post/future-of-pull-requests) ## Masks and Cross-Attention: Extending Self-Attention **Published:** May 3, 2026 **Tags:** AI, Math **Link:** [View post](/post/masks-and-cross-attention) ## Encoder, Decoder, Both: The Three Transformer Architectures Self-attention is the mechanism. BERT, GPT, and T5 are three different ways to stack it. **Published:** May 3, 2026 **Tags:** AI **Link:** [View post](/post/transformer-architectures) ## From Kernel Regression to Self-Attention: A Line-by-Line Derivation **Published:** April 20, 2026 **Tags:** AI, Math **Link:** [View post](/post/kernel-to-attention) ## The Missing Compiler: Automating First-Level Support How I automated first-level support with a workflow, why the state space outgrew it, and what I'd try differently next time. **Published:** April 19, 2026 **Tags:** AI, Customer Support, Architecture **Link:** [View post](/post/workflow-vs-agent) ## Book Review: Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems Practical, critical review of 'Agentic Design Patterns' — evaluates agent design techniques, code examples, and their real-world usefulness for engineers and researchers. **Published:** March 29, 2026 **Tags:** AI **Link:** [View post](/post/agentic-design-book-review) ## Unite x TUM.ai Data Mining Hackathon: Core Demand Prediction How we tackled Challenge 2 at the Unite x TUM.ai Data Mining Hackathon: fee-aware portfolio optimization for recurring procurement demand, from raw PLI data to a two-stage model and NACE-informed cold start. **Published:** March 8, 2026 **Tags:** Data, ML, Hackathon **Link:** [View post](/post/unite-tum-ai-data-mining-hackathon-core-demand) ## Langfuse in Production: Monitoring, Persistent Traces, and a Self-Improving Codebase How I use Langfuse in an AI support pipeline for monitoring, persisting evaluation runs for later review, closing the loop with a multi-agent triage loop driven by Cursor, and a decision framework for when to fix eval data versus prompts. **Published:** February 25, 2026 **Tags:** AI, Python, Observability **Link:** [View post](/post/langfuse-monitoring-eval-self-improvement) ## Type-Safe Hybrid Workflows with Pydantic AI One Pydantic AI primitive scales across deterministic, LLM-assisted, and agentic steps — with the same typed contract everywhere. **Published:** February 25, 2026 **Tags:** AI, Python **Link:** [View post](/post/pydantic-ai-type-safe-hybrid-workflows) ## Temporal for Human-in-the-Loop: When You Don't Know How Long to Wait How Temporal's durable waits, signals, and deterministic recovery make human-in-the-loop workflows reliable when approval times are unknown. **Published:** February 25, 2026 **Tags:** Software Engineering, Workflows, Temporal, Reliability **Link:** [View post](/post/temporal-human-in-the-loop) ## Strategic Model Selection in Cursor: Balancing Cost and Performance Learn how to optimize your Cursor usage by choosing expensive models for planning and cheap models for execution, with a cost vs performance chart updated daily. **Published:** February 6, 2026 **Tags:** AI, Productivity **Link:** [View post](/post/cursor-model-selection-strategy) ## Orchestrating ML Data Pipelines with Snakemake: A Real-World Case Study How I used Snakemake to manage a complex 19-stage ticket analysis pipeline, leveraging scatter-gather parallelization to efficiently process 2000+ customer support tickets with LLM classification. **Published:** February 6, 2026 **Tags:** Data Engineering, Python **Link:** [View post](/post/snakemake-ticket-analysis-pipeline) ## Prompt Injection I participated in a white hat hacking challenge at Agent Olympics Hackathon Munich 2026. Can you trick my chatbot into revealing the secret password? **Published:** January 24, 2026 **Tags:** AI, Security **Link:** [View post](/post/prompt-injection) ## Stop Prompting, Start Programming with DSPy DSPy is a framework that separates the 'what' from the 'how' in LLM development, moving from brittle prompt engineering to robust software engineering. **Published:** January 2, 2026 **Tags:** AI **Link:** [View post](/post/stop-prompting-start-programming-dspy) ## Useful Resources for AI Agents A curated collection of essential resources for building and deploying AI agents, covering frameworks, best practices, and engineering insights from leading organizations. **Published:** January 2, 2026 **Tags:** AI, Resources **Link:** [View post](/post/useful-resources-for-ai-agents) ## Building My Personal Website: Technical Highlights A deep dive into the technical implementation of my personal website, covering the AI chat interface, markdown context generation, and resume PDF sync from a dedicated CV repo. **Published:** December 7, 2025 **Tags:** Web Development, AI **Link:** [View post](/post/building-my-personal-website) ## Extending HDP for Mutational Signatures: Incorporating Evolutionary Trees An extension of the Hierarchical Dirichlet Process model that incorporates phylogenetic tree structures to capture evolutionary relationships in mutational signature activities. **Published:** September 20, 2024 **Tags:** Science **Link:** [View post](/post/mapping-mutational-signatures-to-trees) ## Mutational signature estimation with Hierarchical Dirichlet Process A mathematically rigorous introduction to mutational signature estimation using Hierarchical Dirichlet Processes, bridging the gap between biological applications and statistical methodology. **Published:** September 15, 2024 **Tags:** Science **Link:** [View post](/post/mutational-signature-with-hierarchical-dirichlet-process) ## Deep Learning-Based Diabetic Retinopathy Detection in Fundus Images A deep learning approach for diabetic retinopathy detection using fundus images, combining binary classification and image segmentation to identify disease presence and quantify key indicators. **Published:** May 1, 2024 **Tags:** Science, AI **Link:** [View post](/post/diabetic-retinopathy-detection)