LEARN. BUILD. PROGRESS.

TOPICS

• Computational thinking, algorithmic complexity and Big O notation
• Low-level memory management, pointers, and memory layout in C
• Data structures: linked lists, hash tables, tries, binary search trees
• Algorithms: linear and binary search, sorting (merge, selection, bubble)
• Web development, HTTP protocols, API integration, and SQL relational databases
• Software engineering with Python, Flask, and deterministic test harnesses
• Ten problem sets spanning low-level and high-level systems, plus a comprehensive final project

LEARNED

Working through CS50x from manual byte manipulation, pointers, and heap allocations in C to full-stack systems with Python and relational SQL established a rigorous, first-principles foundation for high-performance software engineering and algorithmic analysis.

ASSESSMENT

Certificate of Completion issued by CS50, Harvard University, awarded by David J. Malan for completing ten problem sets and a capstone project.

VALIDITY

Does not expire

TOPICS

• Data science lab — applied, project-based completion programme

LEARNED

Working through the lab end to end turned data into an analysis that had to stand on its own evidence — every step producing something inspectable, not just described.

ASSESSMENT

Certificate of Completion issued by WorldQuant University.

TOPICS

• Generative AI — internship programme

LEARNED

Building with generative AI in an internship format — the difference between prompting a model and shipping something around one.

ASSESSMENT

Certificate of completion issued for the internship programme.

TOPICS

• US equities fundamentals (US Equities 101)
• Algorithmic trading boot camp track

LEARNED

The boot camp connected the market-data and backtesting work I already do to how US equity sessions and orders actually fit together.

ASSESSMENT

Certificate of completion issued for the boot camp.

TOPICS

• AI data engineering — 210-hour certificate programme

LEARNED

The 210-hour programme put structure around data engineering for AI systems: pipelines designed for models, not just reports.

ASSESSMENT

Certificate of Completion for the 210-hour online skilling course.

TOPICS

• Generative AI
• Artificial intelligence and machine learning
• Aligned to QP SSC/Q8108 · NSQF level 7

LEARNED

Fifteen hours structured around what generative AI is actually doing under the hood — the machine-learning foundations that keep the hype honest.

ASSESSMENT

Certificate of Participation for the 15-hour online skilling course.

TOPICS

• AI DevOps — online skilling course

LEARNED

It gave me the vocabulary of AI delivery pipelines — the names for things I was already assembling by hand.

ASSESSMENT

Certificate of Participation for the 1-hour online skilling course.

TOPICS

• National Occupational Standards — Data Engineering and Analysis (SSC/Q1005), version 4.0

LEARNED

Studying against a published occupational standard gave me a fixed reference point: data work measured against what the standard defines, not against what happens to run.

ASSESSMENT

Certificate of Achievement issued for completing the standard.

TOPICS

• Python
• Data structures and algorithms

LEARNED

Revisiting data structures and algorithms in Python made the everyday implementation choices — when a dict, when a list, when a heap — deliberate instead of habitual.

ASSESSMENT

Course completion certificate for the Python DSA track.

FOCUS

Building backtest harnesses whose outputs are reproducible run to run, and understanding the limits of what reproducibility proves.

METHOD

• Read the primary literature on backtest overfitting and multiple-comparison correction.
• Implement the harness, then deliberately break determinism and observe which false results appear.
• Record negative results rather than deleting the configurations that produced them.

FOCUS

Treating information-flow correctness as an infrastructure property rather than a discipline the researcher must remember.

METHOD

• Audit every read path in the data layer for reachable future information.
• Write adversarial tests that attempt to extract the future rather than tests that confirm correct behaviour.
• Compare exact decimal storage against float storage through aggregation and joins.

FOCUS

Moving risk controls before order construction so that caps and floors define the trade population instead of annotating it.

METHOD

• Derive position size from stop distance and the risk cap so the cap is structural.
• Model fees and slippage explicitly and filter setups on a cost budget before execution.
• Design the ablation study that would attribute contribution to each individual control.

FOCUS

Learning low-level throughput engineering: memory-mapped I/O, eliminating serialization, and measuring honestly on a published harness.

METHOD

• Build zero-dependency tools in C11 against the POSIX API rather than reaching for a framework.
• Project memory address pages over input to read records in place.
• Publish the benchmark methodology alongside the number so the context travels with the claim.

FOCUS

Designing experiment workflows where nothing reaches promotion without a verified artifact and a documented limitation.

METHOD

• Fix the sequence: hypothesis, experiment, verification, artifact, report, promotion.
• Make the gate structural so a claim cannot be argued past verification.
• Test the gate's own integrity, not only the experiments that pass through it.

FOCUS

Understanding where binary floating point corrupts market data and making that trade-off deliberately rather than by default.

METHOD

• Trace representation error through ingestion, storage, aggregation and joins.
• Quantify the throughput cost of exact decimal arithmetic at research scale.
• Decide the trade-off on the basis of the workload rather than convention.