Pengyi Interview Team
Back to all companies

Company 001 / Quant trading

Jane Street

The PIT North Star for collaborative reasoning, technical precision and research-to-production judgment. Current simulations are evidence-bounded training models, not claims about a guaranteed hiring sequence.

Active simulations

Package 004 A-H

Eight role-specific simulations / 32 original rounds / current status: designed, Human runs unmeasured

004A

Trading breaks to reliable desk action

Published TDOE sequence: take-home exercise, problem-solving Zoom, technical Zoom and a multi-desk final stage. Includes original case files, answer structures, code, timer, notes and scoring.

  • Settlement
  • Sanity checks
  • Reconciliation
  • Prioritization
  • Counterparties
Eligibility gate Open TDOE simulation →
004B

Manual controls to reliable infrastructure

Workflow discovery, false-green incident triage, executable freshness controls and a staged global spreadsheet migration. Includes code, tests, timer, hidden answers, notes and scoring.

  • Spreadsheets
  • Freshness gates
  • Idempotency
  • Incident response
  • Migration controls
Experienced bridge Open Infra simulation →
004C

Messy market data to reliable research input

Recruiter alignment, peer problem solving, senior evidence defense and a cross-functional final panel. Includes timer, hidden answers, notes, scoring and local progress.

  • Point-in-time data
  • Lineage
  • Reconciliation
  • Schema drift
  • Human authority
Yellow gate Open Data simulation →
004D

Noisy data to falsifiable research

Research validity audit, executable point-in-time experiment, cost-aware model selection and a cross-functional signal review. Includes Python tests, timer, hidden answers, notes and scoring.

  • Leakage
  • Walk-forward
  • Baselines
  • Costs
  • Productionization
North Star stretch Open QR simulation →
004E

Ambiguity to firmwide action

Published-shape SP practice covering a quantitative online assessment, new-information updates, technical system flows and a final strategic decision with explicit opportunity cost.

  • Problem framing
  • Information flow
  • Prioritization
  • Stakeholders
  • Update speed
2028 eligibility gate Open SP simulation →
004F

Experiment to reliable ML system

Train-to-serve contract design, reproducibility debugging, end-to-end p99 profiling and a bounded shadow decision. Includes a manifest validator, conservative release gate and four new scenes.

  • Reproducibility
  • Train/serve parity
  • Profiling
  • Tail latency
  • Shadow gates
North Star / scale gap Open MLE simulation →
004G

Empirical result to research decision

Claim validity, budgeted baselines and ablations, ambiguous-result interpretation and a continue/pivot/stop panel. Includes paired- result code and explicit publication-depth boundaries.

  • Leakage
  • Baselines
  • Ablations
  • Uncertainty
  • Research taste
North Star / evidence gap Open MLR simulation →
004H

Clarification to collaborative correction

One original event-stream problem grows through implementation, idempotency, API/state design and code review. Uses real Python, focused tests and Jane Street's published collaborative SWE principles.

  • Real code
  • State machines
  • Idempotency
  • Concurrency
  • Code review
Stretch / coding gate Open SWE simulation →

Package 004

Role branches

Internal evidence maps. Not eight simultaneous applications.

A

TDOE

Active mock

B

Infrastructure Automation

Active mock

C

Data Engineer

Active mock

D

Quantitative Research

Active mock

E

Strategy & Product

Active mock / 2028 gate

F

ML Engineer

Active mock / scale gate

G

ML Researcher

Active mock / evidence gate

H

Software Engineer

Active mock / coding gate

Process boundary

Official signal, bounded inference.

Jane Street states that interview structure varies by discipline and can be bespoke. PIT models four pressure levels for practice; it does not label the final round a CEO interview unless a recruiter confirms that format.

Official interview guidance →