PROFILE / ABOUT

Akshay Bande

AI Engineer · Quantitative Research · Algorithmic Trading · C++ Systems

I am an AI and quantitative-finance engineer working at the intersection of machine learning, financial markets and high-performance software.

My background combines more than six years of software and financial-systems engineering with an MSc in Artificial Intelligence from Maastricht University. My current work focuses on quantitative research, financial time series, systematic trading, market microstructure and performance-oriented C++ systems.

Beast's Almanack is where I document the research, systems, strategies, experiments and lessons behind that work.

View Resume GitHub Contact Me

01 / EDUCATION

Education

Maastricht University

Maastricht, The Netherlands

Sep 2024 – Aug 2026

Master of Science in Artificial Intelligence

Research: Financial time-series analysis, Matrix Profile, LoCoMotif, Hidden Markov Models, volatility regimes, motif discovery, statistical testing, crypto, equities, FX and volatility markets.

Selected Coursework

Advanced Machine Learning
Reinforcement Learning
Advanced Natural Language Processing
Intelligent Search and Games
Agents and Multi-Agent Systems
Data Mining
Knowledge Graphs
Autonomous Robotic Systems

University of Mumbai

Mumbai, India

Jun 2014 – May 2018

Bachelor of Engineering in Computer Engineering

Selected Coursework

Data Structures and Algorithms
Computer Architecture
Probability
Linear Algebra
Discrete Mathematics
Database Management Systems
Computer Networks
Computational Theory
Data Mining

02 / EXPERIENCE

Professional Experience

Catella Investment Management

Maastricht, The Netherlands

Jul 2025 – Dec 2025

Intern Data Scientist / Quantitative Investment Research

Conducted quantitative investment research linking ESG indicators, financial fundamentals and macroeconomic variables to portfolio risk, capital allocation and sustainable investment decisions.
Developed client clustering and segmentation models using unsupervised machine learning to identify investor profiles, portfolio characteristics and heterogeneous client groups.
Built Python-based financial-data workflows covering ingestion, cleaning, feature engineering, exploratory analysis, statistical modeling and investment research.
Worked with multi-source financial datasets to investigate relationships between investment characteristics, climate-risk variables and portfolio-level outcomes.

Volkswagen Group Technology Solutions

Pune, India

Mar 2023 – Jul 2024

Senior Developer, Enterprise Systems

Developed high-throughput enterprise data systems supporting global vehicle operations using SAP HANA, ABAP, CDS Views and SAP S/4HANA.
Engineered performance-sensitive database logic and data pipelines with emphasis on query latency, system throughput, reliability and large-scale transaction processing.
Implemented SAP Extended Warehouse Management solutions supporting Audi and CUPRA warehouse operations for Volkswagen Australia.
Contributed to migration and integration of legacy enterprise systems into SAP S/4HANA while maintaining production continuity and data integrity.
Mentored engineers in system architecture, backend development and performance-oriented implementation.

Accenture

Mumbai, India

Sep 2021 – Mar 2023

Developer, Financial and Enterprise Systems

Developed and maintained enterprise financial systems and transaction-processing logic for large-scale international clients.
Supported brownfield SAP implementation and migration initiatives for Maserati under Stellantis Group and developed enterprise solutions for Ferrero.
Designed backend application logic, database workflows, APIs and integration components for business-critical financial and operational systems.

Invenio LSI

Manama, Bahrain

Oct 2018 – Sep 2021

Developer, Financial Systems

Designed and developed financial calculation, registration and transaction-processing components for the Kingdom of Bahrain's national tax platform.
Contributed to government financial systems for Saudi Arabia's Zakat authority, Qatar General Tax Authority and Fiji Revenue and Customs Service.
Developed business-critical financial logic supporting high-volume taxpayer, registration and transaction workflows across GCC taxation platforms.
Received the Best Performer award for contributions to the Bahrain Integrated Tax System and mentored 10+ developers in backend and rules-engine development.

03 / TECHNICAL EDGE

Where My Work Converges

AI & Machine Learning

Machine learning, representation learning, deep learning, Hidden Markov Models, reinforcement learning and intelligent systems.

Quantitative Research

Financial time series, statistical modeling, motif discovery, regime analysis, Monte Carlo methods and reproducible experimentation.

Trading & Market Systems

Market microstructure, bid/ask data, transaction costs, slippage, execution simulation, backtesting and systematic strategies.

Performance Engineering

C++, concurrency, multithreading, algorithms, memory management, Linux, networking and performance-sensitive software.

04 / SKILLS

Technical Arsenal

Programming

C++
Python
SQL
Bash
JavaScript
Rust
Git
Linux

C++ / Systems

STL
Memory Management
Concurrency
Multithreading
Templates
Data Structures
Algorithms
Performance Optimization
Networking Fundamentals
TCP/IP
UDP

Python / Scientific Computing

NumPy
Pandas
SciPy
Polars
Scikit-learn
PyTorch
Statsmodels
Matplotlib
STUMPY

Quantitative Finance

Financial Time Series
Statistical Modeling
Signal Generation
Backtesting
Portfolio Analysis
Risk Modeling
Monte Carlo Simulation
Bootstrap Analysis
Factor Analysis
Transaction-Cost Modeling

Algorithmic Trading

Market Microstructure
Bid/Ask Data
Spreads
Slippage
Market Orders
Limit Orders
Stop Orders
Execution Simulation
Position Sizing
Systematic Strategies
Regime-Conditioned Trading

Machine Learning

Regression
Classification
Clustering
Hidden Markov Models
Reinforcement Learning
Neural Networks
CNN
RNN
LSTM
Transformers
Representation Learning

Data Engineering / MLOps

ETL Pipelines
FastAPI
REST APIs
Docker
CI/CD
Data Validation
Experiment Pipelines
Model Deployment
AWS
GCP

Enterprise Data Systems

SAP HANA
ABAP
CDS Views
OData
SAP S/4HANA
System Integration
Database Optimization

05 / RESEARCH

Research

RESEARCH 01

2025 – 2026 · Master's Research

Multivariate Motif Discovery in Financial Time Series

Regime-Agnostic and Regime-Conditioned Methods Under Nonstationarity

Developed a large-scale research pipeline for discovering recurring multivariate patterns in nonstationary financial time series using Matrix Profile and LoCoMotif.

Compared regime-agnostic and regime-conditioned discovery using volatility-quantile regimes and Gaussian Hidden Markov Models.

Evaluated BTCUSDT, ETHUSDT, S&P 500, DAX, NASDAQ-100, EURUSD, GBPUSD and VIX across frequencies from 1-minute to daily data.

Financial Time Series
Matrix Profile
LoCoMotif
HMM
Python
Statistical Testing

RESEARCH 02

Motif Discovery: Quantitative Benchmark & Event Study

Built a reproducible financial-time-series motif benchmark using Matrix Profile, MSTUMP, LoCoMotif and volatility regimes.

BTCUSDT motif observations: 39,906
Strongest regime-agnostic nearest-neighbour distance: D = 0.082
LoCoMotif occurrences across three controlled market slices: 73
Temporal coverage: 69.2–75.2%
Monte Carlo resamples: 2,000
State-dependent 24-hour post-motif effects: +2.73 pp / −10.70 pp

Matrix Profile
MSTUMP
LoCoMotif
Monte Carlo
Event Study

RESEARCH 03

2025 · Catella Investment Management

European Climate-Risk Clustering

Developed a European climate-risk clustering pipeline combining hazard, exposure and vulnerability indicators.

Validated cluster structures against the RESIN expert typology across 1,304 European regions. Created 8- and 10-cluster regional typologies and explained cluster drivers using SHAP.

European regions: 1,304
ARI: 0.2112
NMI: 0.3187

Clustering
Climate Risk
SHAP
Unsupervised Learning
Investment Research

06 / SELECTED WORK

Selected Projects

PROJECT 01

Systematic FX Research & Backtesting Platform

End-to-end FX research with Dukascopy bid/ask data: ingestion, validation, canonicalization and multi-timeframe resampling. Modeled spreads, slippage, costs and execution, with Monte Carlo, bootstrap, regime, cost-stress and ablation analysis.

Python
FX
Backtesting
Market Microstructure
Monte Carlo

PROJECT 02

RB Confluence Systematic Trading Strategy

Multi-signal FX framework combining trend, momentum, volatility, price action, market structure and session information, with rule-based signal scoring, entry logic, stop-loss methodology and execution constraints.

Systematic Trading
FX
Signal Research
Execution

PROJECT 03

Cross-Platform Undisclosed Advertisement Detection

Analyzed 292,315 posts from 150 Dutch influencers using FLAN-T5-XL, 384-dimensional embeddings, Positive-Unlabeled Learning, Active Learning, Logistic Regression and SVM.

LLM accuracy: 88%
PU-SVM accuracy: 90%

NLP
LLM
PU Learning
SVM

PROJECT 04

Quantum-Computing Research Trend Analysis

Analyzed approximately 4,000 arXiv papers (2014–2024) with GPT-4 Turbo, KeyBERT, SciBERT, UMAP, HDBSCAN and BERTopic. Identified Error Mitigation as the fastest-growing modeled subfield.

Projected growth, 2022–2026: +160%

NLP
SciBERT
BERTopic
Forecasting

PROJECT 05

Deep Reinforcement Learning for Autonomous Lunar Landing

Implemented A2C, DDPG and TD3. Full TD3 achieved 12.8% and 14.4% higher final average rewards than the specified reduced TD3 variants in ablation experiments.

Best full-TD3 average reward: 267.8
Training episodes: 1,500

Reinforcement Learning
PyTorch
TD3

PROJECT 06

Autonomous Multi-Robot Exploration

Four coordinated robots with differential-drive kinematics, 24-ray sensing, EKF localization, collision avoidance and log-odds mapping. Genetic Algorithms used 30 genomes over 30 generations; 3,100+ cells discovered within 150 steps.

Unexplored cells in ~45 seconds: <10%
Simulation: 60 FPS

Robotics
EKF
Genetic Algorithms
Python

PROJECT 07

Biodiversity Knowledge Graph

Integrated species-occurrence and conservation data across 8 conservation categories in an ontology/RDF pipeline using Darwin Core, GraphDB and SPARQL.

GBIF occurrence records: 500K+
IUCN Red List records: 5,895

Knowledge Graphs
RDF
SPARQL
GraphDB

PROJECT 08

Fianco Game AI

Complete 9×9 adversarial game AI using Negamax, Alpha-Beta pruning, iterative deepening, quiescence search, killer-move and history heuristics, transposition tables and 64-bit Zobrist hashing.

AI decision window: 5 seconds

Python
Adversarial Search
Game AI
Algorithms

07 / RECOGNITION

Awards & Recognition

Best Performer Award

Bahrain Integrated Tax System

Recognized for development of core registration and agent-management systems supporting Bahrain's national tax implementation.

Technical Leadership

Mentored 10+ engineers across Invenio LSI and Volkswagen in software development, architecture and performance-oriented implementation.

Volkswagen Group

Recognized for contributions to SAP EWM implementation supporting Audi and CUPRA warehouse operations in Australia.

08 / CERTIFICATIONS

Certifications

Technology

SAP Certified Development Specialist
SAP Certified Activate Project Manager

Languages

IELTS Academic — Overall Band 8.0
Dutch — A1 Certification

Beyond Work

Quantitative Trading · Financial Markets · Competitive Dota · C++ Systems Engineering · Artificial Intelligence · Strength Training · Reading · Market Psychology

LET'S CONNECT

Interested in AI, markets or high-performance systems?

I am particularly interested in quantitative research, algorithmic trading, AI/ML engineering and performance-oriented software systems.

Contact Me

RESUME

Want the full picture?

Download my resume for a concise overview of my experience, research, engineering projects, education and technical background.

Download Resume