I'm Matt.

I am a Senior Machine Learning Engineer based in Helsinki, currently focusing on building scalable lead generation engines at KONE, optimizing MLOps infrastructure and building beautiful smooth pipeline between Databricks and AWS.

With over 10 years of experience in the intersection of Machine Learning, Data Engineering, software architecture, I specialize in turning complex algorithms into production-ready distributed systems. I have a deep love for clean code, robust backend architecture and playing games.

Email LinkedIn GitHub

Work Experience

Senior Machine Learning | Software Engineer @ KONE

2025 — Present
Helsinki, Finland
PythonDatabricksAWSAmazon BedrockAWS LambdaStep FunctionsEventBridgeAppFlowSalesforceReactTypeScriptFastAPICDKML PipelinesGenAI / RAG
  • Technical lead for the Lead Engine, owning the platform's technical direction end to end and driving its productionization and rollout across 70+ countries.
  • Delivered a GenAI document-ingestion pipeline (Microsoft Graph API, AWS Lambda, EventBridge, Amazon Bedrock) that syncs SharePoint knowledge documents into a Bedrock Knowledge Base behind an LLM sales assistant, tagging each document with country-level access metadata so retrieval respects user entitlements; shipped with 100% Lambda test coverage and three-environment CI/CD.
  • Built an end-to-end AI lead-generation pipeline for service contracts that discovers equipment under contract, enriches it from Salesforce and internal equipment and service data, and creates CRM-ready leads with approval-controlled dispatch, taking the first campaign from prototype to production.
  • Designed and built Lead Hub, the full-stack control room for the Lead Engine (React/TypeScript, FastAPI on AWS Lambda, Databricks, AWS CDK), where business owners tune the ~300 configurations that decide which leads are generated and how an ML win-probability model is weighted against rule-based scores, with versioned, audited edits and country-level access control.
  • Received a KONE invention reward for an ML delivery-risk tool that predicts construction-milestone delays (risk classification, quantile regression), so DACH operations can prioritize at-risk projects early; owned the full stack from data engineering to a scored dashboard app on AWS.
  • Engineered the Lead Engine pipeline that turns internal and external data into ranked Salesforce leads, loading five Salesforce objects from S3 via AWS AppFlow, Step Functions and EventBridge with automated dev/QA/prod deployments.

Machine Learning | Software Engineer @ Resoniks

2024 — 2025
Helsinki, Finland
PythonGrafanaMLflowPrometheusAirflowPyTorchTransfer LearningFastAPI
  • Designed and implemented a scalable MLOps infrastructure using Grafana, MLflow, Prometheus, and Airflow, reducing model deployment time by 40% and improving monitoring efficiency by 30%.
  • Developed and optimized backend services for deploying and managing machine learning models in production, ensuring 99.9% uptime and reducing response time by 25%.
  • Engineered and integrated an autoencoder-based anomaly detection system, reducing false positives by 10%, enhancing system reliability, and improving real-time detection accuracy.
  • Leveraged transfer learning to fine-tune pre-trained models for audio anomaly detection, improving accuracy by 10% with limited training data and reducing computational costs by 20%.
  • Built and deployed production-grade services for data export handling, automating workflows with Airflow, reducing manual intervention by 50%, and increasing data processing speed by 2x.

Software Engineer @ DGShahr

2022 — 2023
Tehran, Iran
PythonDjangoDRFCeleryRedisPostgreSQLREST APIs
  • Built a scalable and modular backend architecture with Django, designing custom models, views, and APIs, increasing data processing efficiency by 30%.
  • Improved task processing speed by 15% by implementing asynchronous operations with Celery and Redis, ensuring faster execution of critical workflows and reducing system bottlenecks.
  • Developed and optimized REST APIs using Django REST Framework (DRF), improving data retrieval performance by 30% and enabling seamless integration with front-end applications.
  • Optimized complex queries and enhanced database performance by implementing Django ORM and PostgreSQL time series, reducing query execution time by 20% and improving scalability.

Data Engineer @ Divar

2022
Tehran, Iran
PythonSQLPandasA/B TestingData PipelinesFeature Engineering
  • Developed automated reporting pipelines to assess fraud detection algorithm performance, revealing a 10% reduction in financial losses and improving fraud mitigation strategies.
  • Improved detection of fake reporters by 15% through advanced data analysis, feature engineering, and algorithm optimization, leading to more reliable fraud detection.
  • Designed and conducted A/B tests to analyze the impact of the new KYC process on user engagement, identifying a 20% churn rate and providing data-driven insights for decision-making.
  • Engineered and optimized backend data pipelines for fraud detection using Python and SQL, enhancing system efficiency and reducing query execution time by 25%.

Data Engineer @ Hesaba

2020 — 2022
Tehran, Iran
PythonScalaFastAPIPostgreSQLRedisKafkaFaissAnnoyMinHashMicroservicesRecommender Systems
  • Developed high-performance similarity detection algorithms for CDR (Call Detail Record) big data, achieving 80% accuracy using Jaccard Index, Annoy, Faiss, and MinHash.
  • Conducted data-driven research and model evaluations, refining AI-driven solutions for telecom analytics, geospatial intelligence, and anomaly detection.
  • Engineered scalable ML pipelines for large-scale graph-based and nearest-neighbor search algorithms, optimizing computational efficiency for billions of records.
  • Utilized Scala for real-time data streaming and processing, optimizing pipeline efficiency for telecom data analysis.
  • Designed and implemented scalable microservices and RESTful APIs for a personalized recommendation system, integrating autoencoders using Python and FastAPI, optimizing data processing pipelines, and improving conversion rates by 10%.
  • Developed and optimized backend services using PostgreSQL, Redis, and message queues (Kafka) for recommendation pipelines, implementing collaborative and content-based filtering models to serve real-time personalized product suggestions and recommendations based on user behavior and purchase history.

Education

Aalto University

2023 — 2025
M.Sc. Computer Science

Minor: Artificial Intelligence, Machine Learning and Data Science

G.P.A: 4.56/5 - Pass with Honor grade. Awarded the Dean's Scholarship for academic excellence.

Master's Thesis

Dual-Model Augmentation for Compositional Semantic Parsing via Recursive Log-Likelihood Guidance (Via LLM)

Supervisor: Prof. Pekka Marttinen

Advisor: Sam Spilsbury

Key Courses

Machine Learning5/5
Machine Learning: Supervised Methods5/5
Reinforcement Learning5/5
Deep Learning5/5
Federated Learning5/5
Statistical Natural Language Processing4/5
Bayesian Data Analysis4/5
Principles of Algorithmic Techniques5/5