CV#
Download PDF:
cv.pdf
Konsta Kiirikki#
konsta.kiirikki@gmail.com | 0401439410 | https://www.linkedin.com/in/konsta-kiirikki | PlayerPlanet | https://users.aalto.fi/~kiirikk1
Education#
Bachelor’s Programme in Science and Technology — BSc in progress (2024–) Aalto University · Major: Mathematics and Systems Sciences
Strongest results on the applied / computational, optimisation, time-series and inference courses (Dynamic Optimization, Introduction to Optimization, Prediction & Time Series Analysis, PDEs, and probability all at top marks). Member of the Physics Guild (Fyysikkokilta).
Intended minor in Data Science — planned, not yet on record.
Relevant coursework (selected):
Simulation-based Inference (special course, early 2026 — the method now anchoring EASE, StimIQ, and QSVT4CRA)
Dynamic Optimization (master’s-level, MS-E)
Introduction to Optimization
Prediction and Time Series Analysis
Machine Learning
Partial Differential Equations
Stochastic Processes
Probability and Statistics
Linear Algebra · Matrix Algebra · Differential and Integral Calculus
Course trajectory: physics entry (mechanics, thermodynamics, electromagnetism, engineering-physics calculus) → probability and core mathematics → ML + time-series (autumn 2025, concurrent with the AI4Value role and EASE) → optimisation and inference (early 2026, overlapping StimIQ’s SNPE build). The method was being learned in class and shipped in hackathons in the same months.
Experience#
AI/ML Developer · AI4Value · May 2026 – Present (full-time)
MLOps, algorithm research, anomaly detection. Currently building a Stream Analytics Engine and sklar-engine (dependence-structure modelling — the name nods to Sklar’s theorem / copulas).
Junior Software Developer · AI4Value · Sep 2025 – May 2026 (part-time)
Software development; local LLMs.
Trainee · AI4Value · May 2025 – Sep 2025 (internship)
Software development and AI on a Python + Django stack.
Across the three roles at AI4Value (continuous employment since May 2025), the work spans algorithm modules, Bayesian neural networks, and AI agents. Tools: Python, Docker, CI/CD; cloud deployment (Azure, Verda).
IoT Developer Trainee · University of Oulu · June 2019
Designed and built an ESP32 sensor network with WiFi/Bluetooth, sharing data to end users through a web interface.
Selected projects#
EASE Health — personalised migraine prediction
Amortised Bayesian inference engine (Bayesian neural network) forecasting individual migraine risk and quantifying uncertainty.
Continuously optimises for information gain via active learning.
Trained offline on synthetic data; active-learning convergence demonstrated in deployment.
Gold on the Pfizer × Aava track at Junction Main. Accepted to Spark Finland’s 2026 autumn batch.
StimIQ — Deep Brain Stimulation parameter optimisation
Decision-support system for tuning DBS parameters in Parkinson’s disease.
Dense daily symptom data (PDQ-39) feeds a mechanistic “digital-shadow” forward model; amortised inference (SNPE) + Pareto multi-objective optimisation produce a recommended configuration the clinician reviews and applies — human-in-the-loop by design.
My involvement: original StimIQ built with Atte Laakso, Nikolas Juhava, Niklas Keiski. The EuroTech × Hong Kong revision evolved without me — by Atte Laakso, Marc Chen, Qilun Li, Nikolas Juhava.
Original repo: PlayerPlanet/StimIQ · demo: https://stim-iq.vercel.app · EuroTech × HK track (by others): Latt33/StimIQ
QSVT4CRA — credit / real-estate risk via QSVT on LUMI
Quantum Singular Value Transformation applied to credit and real-estate risk on Finnish data across 17 regions; deployed on the LUMI supercomputer and on IBM quantum hardware (Boston during the hackathon, Heron r3 for the scaling study).
Pipeline: train NPE over regimes → posterior samples for a factor-copula → block-encode the factor-copula parameters into a QSVT unitary → run QAE to estimate VaR + CVaR.
Silver on the OP Pohjola track at Junction X OP Pohjola 2026 (with Milla Kolehmainen, Antti Nurminen, Silja Heiskanen, mentor Juha Vesanto).
Honest result: ~10% VaR-95 tail-estimate error on a 12-asset portfolio on real IBM hardware, scaling with QSVT polynomial degree / circuit depth.
Companion paper: Implementing Credit Risk Analysis with Quantum Singular Value Transformation.
Write-up: Junction X OP Pohjola: QSVT4CRA — silver
MANTIS-4D — pose estimation for the LeRobot SO-101
Depth-based policy model for the SO-101 arm.
Physical AI hackathon submission (Aaltoes, Oct 2025); arxiv idea based on https://arxiv.org/html/2501.08329v1.
Repo: PlayerPlanet/MANTIS-4D
AaltoAI: Predicting Electricity Spot Prices
LSTM on weather and consumption data (TensorFlow, pandas/NumPy).
RSSI-Based Indoor Location System
Device localisation from advertisement-call signal strength, fitting physics-based non-linear models via least-squares.
Coffee level prediction (kahvibot)
Vision neural networks estimating coffee-pot fill level for the Physics Guild’s “kahvibot.”
Repos: PlayerPlanet/coffee_level_detection · PlayerPlanet/coffee_images
Hackathons#
Junction Main — Pfizer × Aava track: EASE, gold
Eurotech × Hong Kong — StimIQ (DBS digital-shadow decision support; track evolved independently after my involvement ended)
Junction X OP Pohjola 2026 — OP Pohjola track: QSVT4CRA, silver (with Milla Kolehmainen, Antti Nurminen, Silja Heiskanen; mentor Juha Vesanto)
Aaltoes RoboHack (Oct 2025) — robotics + AI
Sunstead Hack (Levi, Lapland, June 2026) — agentic AI
AI Buildathon (Make in Finland / DIMECC) — industrial AI with Bluefors, Fortaco, Kesla, Konecranes
Skills#
Hard skills#
Methods: amortised / simulation-based Bayesian inference (SNPE, sbi); mechanistic forward models / digital shadows; dynamical systems; time-series analysis; optimisation & dynamic optimisation; active learning & information-gain–driven experimental design; Pareto multi-objective optimisation; copulas & dependence modelling; stochastic processes; statistical & probabilistic modelling.
Programming: Python, Scala, C++, MATLAB, R
Frameworks & tools: PyTorch, sbi, TensorFlow, pandas/NumPy; Docker, CI/CD; cloud (Azure, Verda) and HPC (LUMI); FastAPI / SQLite and React + TypeScript in collaborative full-stack projects; agentic-coding workflows (Claude Code, OpenCode).
Soft skills#
Mathematical and analytical thinking
Teamwork and cooperation
Design and problem solving