Skip to content

Ilia Duda

Quantitative analyst and engineer at a proprietary options trading firm since January 2026. I build the tools that automate the firm’s investment workflows for its traders: real-time dashboards, analysis tooling, and research tooling around the firm’s proprietary options model. Mathematics and Business Administration at Northeastern, class of 2028.

Seeking
Open to a 6-month co-op (an off-cycle internship) from January 2027
Contact
Roles
Quant and risk · Investments · Investment banking · Data science in finance
Location
Based in Boston, MA; open to New York, San Francisco or London
Fig. 1
Every line is one possible year for a $100 stock; together they price a call.Simulated · geometric Brownian motion · σ 23%, the simulated market’s realized vol · r 3% · 64 steps · not market data
Simulated ± 2 SE
10.55 ± 0.13
Black–Scholes
10.5742
Gap to the formula
0.4 SE
Paths simulated
65,536
Priced
at build time

Click a price at expiry to set the strike · drag sideways for volatility.

Each line is a path of geometric Brownian motion, stepped exactly in log space; its random numbers come from a counter-based hash, so any path can be regenerated on the CPU, where the tests check the estimator against Black–Scholes. A call pays whatever the stock finishes above the strike, and its price is that payoff averaged over every path and discounted to today: the indigo bars, what each ending pays weighted by how often it happens, add up to it. The readouts show the estimate closing in on the formula as the paths pile up. The volatility starts at the simulated market’s own realized volatility at one moment, the one /market opens on, 23.0% (to the slider’s 1% step), computed again by your browser.
Where 65,536 simulated paths of a $100 stock end after one year at 23% volatility, and what a call struck at $100 pays there. Priced by simulation at 10.55 ± 0.13; the Black–Scholes formula gives 10.5742.
Price at expiryShare of pathsAverage payoff there
$20–$400.0%$0.00
$40–$601.2%$0.00
$60–$8015.2%$0.00
$80–$10033.0%$0.00
$100–$12028.8%$9.23
$120–$14014.4%$28.60
$140–$1605.2%$48.18
$160–$1801.5%$67.93
$180–$2000.5%$87.04
$200–$2200.1%$107.93
$220–$2400.0%$128.78

These futures, its order book and its vol surface: one simulated market, running in your browser.

Contents

  1. One market, three views

    Independent work · September 2026 · synthetic data

    One simulated market runs in your browser, drawn three ways in one frame: its order book, a year of futures, its vol surface. A liquidity shock hits all three.

  2. CloseBooks: a multi-tenant month-end close with an LLM in the loop

    Founder and sole engineer · April 2026 – present · synthetic feed

    A multi-tenant month-end close for CPA firms that I built alone: an LLM pipeline maps every bank line to the client’s accounts, and only confident or reviewer-approved lines export.

  3. The vibrating drum: Fourier–Bessel series on a circular membrane

    Directed study · MATH 4992 · Spring 2026

    A drum ringing in the Fourier–Bessel modes the reader keeps, from a directed study’s exercises: Bessel functions and zeros computed in the page with no library, held to SciPy.

  4. What ball-by-ball cricket predicts beyond the scoreboard

    Independent research · July 2026

    A leakage-audited model of T20 cricket, built on 4,748,382 T20 and ODI deliveries: gradient boosting on match state cuts win-probability log-loss 29% below the base rate on held-out matches.

  5. Order flow that remembers: a Hawkes-driven limit order book

    Independent work · September 2026 · synthetic data

    A synthetic limit order book driven by a six-kind Hawkes process, simulated exactly in your browser and drawn as terrain: read the odds of what set off any market order.

  6. An implied-volatility surface free of static arbitrage

    Independent work · September 2026 · synthetic data

    A synthetic SSVI volatility surface in live 3D, with local vol and Black–Scholes Greeks at any point, that takes a shock and stays free of static arbitrage in every frame.

  7. Ranking startup segments, and how much the answer depends on the data

    Data-analytics capstone, Yandex Practicum · December 2025 · Python, pandas

    A composite model ranks 48 startup segments on growth and size; changing one data decision at a time gives three treatments three top picks, and shows which segments hold up.

  8. A debt-settlement portal built to Russian federal law

    Sole developer and project lead · July 2026 – present · illustrative terms

    A self-service portal for settling a debt without a phone call, built end to end as sole developer for a licensed Russian collection organization, with federal law enforced in code.

Experience

  • Glacier Capital Systems, proprietary options trading firm

    Quantitative analyst and engineer · Remote · January 2026 – present

    Building the tools that automate the firm’s investment workflows for its traders: real-time dashboards, analysis tooling, and research tooling around the firm’s proprietary options model.

  • Debt-settlement portal, licensed Russian collection organization

    Sole developer and project lead · Remote · July 2026 – present

    Building a regulated consumer product end to end: debt lookup, a settlement calculator, SMS authentication and an SBP payment screen, under 152-FZ and 230-FZ. Read the paper

  • CloseBooks, month-end close for accounting firms

    Founder · April 2026 – present

    Built and deployed a multi-tenant month-end close product for CPA firms on my own, with an AI categorization pipeline on the Claude API. Read the paper

  • BCS Bank, Investment Banking Division

    Investment banking intern · Moscow · July – August 2023

    Covered Russian energy, metals and banking; built DCF, comparable-company and sensitivity models; wrote daily briefings on government bonds through two rate rises. See the curve

  • Monito, Young Enterprise UK

    Co-founder and financial director · UK · 2022 – 2023

    Won UK National Company of the Year and reached the European Finals.

Every role in full, with education and courseworkThe one-page CV

Other work

  • nucarbonDeveloper · May 2026 · Northeastern Sustainability Incubator · deployedA dashboard estimating the carbon cost of AI use across a university campus
  • AdConfirmCo-founder · May 2026 – presentAdvertising inside invoices and receipts, across eight accounting and point-of-sale integrations, with metered billing