Jane Street
Showing 61–75 of 117 transcripts.
- 1h 3m
An Inside Look at Jane Street’s Tech Internship with Jeanne Van Briesen, Matt Else, and Grace Zhang
Jeanne Van Briesen, Matt Else, Grace Zhang, Ron Minsky
Jane Street's internship program recruits diverse candidates without requiring prior OCaml knowledge, grounding them in a boot camp before assigning them to dual-team rotations on production-critical infrastructure. Recent projects by interns such as Grace Yang and Matt Els have directly implemented risk management features and redesigned API systems, while a rigorous code review culture ensures these contributions are safely integrated into live environments. The transition from intern to full-time engineer involves navigating a steep technical learning curve and balancing feedback dynamics within a community that actively pursues demographic diversity across non-traditional feeder schools.
- 1h 4m
Building a Functional Email Server with Dominick LoBraico
Faced with the security risks and operational rigidity of a legacy mail server, Jane Street engineers replaced the system with MailCore, a homegrown server implemented in OCaml where configuration is expressed as composable code. To ensure safety during the one-year migration, the team deployed a parallel shadow instance that continuously diffed outputs against the legacy system, effectively reverse-engineering implicit knowledge while eliminating memory-safety bugs common in C-based alternatives. This architectural shift transformed email management into a software culture practice, allowing generalist engineers to safely modify routing and filtering logic using standard OCaml tools rather than relying on specialized knowledge of opaque custom domains.
- 1h 8m
Language Design with Leo White
Jane Street compiler team member Leo White discusses OCaml's unique design philosophy, which balances pure mathematical type theory with practical modularity to optimize for compiler construction. He outlines current limitations in memory layout and dependent types that hinder systems programming, while highlighting upcoming features like modular implicits and algebraic effects aimed at resolving compositionality issues without sacrificing safety. The conversation further details Jane Street's collaborative upstream strategy, which uses internal forks to experiment with risky innovations while maintaining rigorous backward compatibility for the broader open-source community.
- 45 min
Clock Synchronization with Chris Perl
Responding to MIFID 2 regulations, Jane Street engineered a hybrid time synchronization system that combines GPS-driven Precision Time Protocol with Linux-based NTP Interleaved Mode to achieve sub-100-microsecond accuracy across its infrastructure. By leveraging hardware timestamping at the network interface and cross-verifying local and distant time sources, the architecture mitigates standard PTP vulnerabilities while maintaining a simple, inspectable network topology. This design successfully reduces worst-case synchronization errors to approximately 35 microseconds, effectively balancing regulatory compliance with robust fault tolerance.
- 1h 0m
Python, OCaml, and Machine Learning with Laurent Mazare
Jane Street employs a dual-language architecture that leverages OCaml for robust, statically-typed production systems while utilizing Python for rapid machine learning research and data analysis. Laurent Mazarin highlights how the firm bridges these environments through the PyML library to enable bidirectional communication, though this integration faces technical challenges regarding garbage collection and memory management. To address evolving needs, the organization is exploring Swift for automatic differentiation and investigating algebraic effects in OCaml to achieve fine-grained resource control comparable to Rust.
- 1h 10m
Compiler Optimization with Greta Yorsh
Jane Street compiler engineer Greta Jorsch presents a strategic shift toward combining formal verification with targeted testing and super optimization to address the undecidability limits of pure verification in compiler design. Her team's implementation of sampling-based Feedback-Directed Optimization in the OCaml compiler achieved 10–20% performance gains by stabilizing runtime behavior and minimizing cache misses while managing the trade-off between portability and architecture-specific micro-optimizations. Jorsch advocates for open-sourcing these performance tooling benchmarks to influence upstream development, emphasizing that modular compiler passes can balance internal innovation with community maintainability.
- 1h 2m
Multicast and the Markets with Brian Nigito
Over the past two decades, securities exchanges have transitioned from physical floors to highly regulated, electronic ecosystems where ultra-low latency determines competitive advantage. To achieve this, firms utilize specialized networking architectures like UDP multicast and co-located hardware to distribute market data faster than standard internet protocols allow, while relying on custom recovery layers and sequencer models to ensure transactional reliability. As hardware speeds approach 25 Gbps, the industry is increasingly shifting from general-purpose servers to specialized FPGAs and custom network cards to manage serialization delays and maintain deterministic performance.
- 58 min
Build Systems with Andrey Mokhov
This presentation analyzes the historical evolution of build systems, highlighting the enduring dominance of `Make` and its specific technical limitations regarding scalability and complexity, as illustrated by the GHC migration to Hadrian. It introduces a theoretical framework that decomposes build architectures into schedulers and rebuilders to propose innovations like Cloud Shake, while examining industry-specific implementations at Jane Street, such as the Dune and Jenga systems. The discussion concludes by exploring emerging trends toward fine-grained incrementality, IDE integration, and the broader convergence of build tools with general-purpose incremental computation models.
- 59 min
Programmable Hardware with Andy Ray
Jane Street leverages FPGA and ASIC hardware to achieve deterministic, line-rate processing for high-frequency financial trading, overcoming the latency and scalability limitations inherent in general-purpose software. To modernize this engineering workflow, the firm utilizes HardCAML, an open-source domain-specific language built on OCaml that enables composable, test-driven hardware design comparable to modern software development. By releasing specialized libraries and realistic examples, Jane Street aims to lower the barrier to entry for complex hardware engineering and accelerate industry-wide adoption of rigorous, software-style verification practices in chip design.
- 12 min
Real Numbers – Episode 16, Finale
The final episode of Season 1 of *Real Numbers* derives five distinct mathematical solutions to the geometric distribution problem of calculating the expected number of half-court shots Danielle needs to make. The episode further analyzes a bonus problem involving two consecutive successes, demonstrating via a recursive state method that the expected attempts equal 30 rather than the intuitive estimate of 25. Concluding the season, the host solicits listener feedback on problem topics, format changes, and potential mathematical depth for the upcoming second season.
- 11 min
Real Numbers – Episode 15, The Half Court Shot
Host discusses probability theory by solving a problem where a student named Jayden schedules two or three exams across five weekdays, calculating expected spans of 1.6 and 2.4 days respectively through enumeration and linearity of expectation. The episode transitions to a new challenge involving a basketball player named Danielle who makes 20% of her half-court shots, asking listeners to determine the expected attempts needed for a single success and for two consecutive makes. While the mathematical framework for the exam scenarios is fully derived and verified, the solution to Danielle's geometric distribution problem remains pending for the listener to solve.
- 6 min
Real Numbers – Episode 14, Finals Week
In this Real Numbers episode, the host explores expected value applications by challenging listeners to calculate the average interval between randomly scheduled final exams and reviewing a previous problem on the expected number of distinct snowboards used over six days. The snowboard scenario is resolved by demonstrating two distinct approaches using indicator variables and the linearity of expectation, both confirming that a user will sharpen approximately 3.99 different boards on average. The episode concludes by inviting the audience to submit their own solutions and alternative problem ideas for future episodes focused on mathematical expectation.
- 6 min
Real Numbers – Episode 13, How Many Snowboards Need Sharpening
This instructional session applies the linearity of expectation to calculate expected values in scenarios involving both independent and dependent selections. Using the marching band example, the analysis demonstrates how fixed global distributions alter probabilities while the method for summing indicator variables remains valid regardless of independence. The presentation concludes by challenging the audience to utilize this specific probabilistic framework to determine the expected number of snowboards requiring sharpening in a student-led lesson environment.
- 10 min
Real Numbers – Episode 12, Photographing the Chaotic Marching Band
The episode explores expected value and the linearity of expectation by first applying these concepts to a conditional probability problem involving a haunted forest monster. The instructors update monster type probabilities based on a 4-damage roll to calculate a resulting expected value of 5.5 for the second attack. The session concludes by introducing a problem of the week that contrasts independent choice scenarios with fixed pool distributions to determine the expected number of synchronized rows in a marching band.
- 10 min
Real Numbers – Episode 11, The Unknown Monster
The episode calculates the expected damage of a random monster attack by weighting uniform distributions across three difficulty tiers and determines the conditional expectation of a second attack following a four-damage outcome. A prior review solves the geometric optimization of a painting passing through a composite window, utilizing the Power of a Point theorem to prove the maximum height corresponds to the Golden Ratio. Both segments demonstrate how probabilistic weighting and geometric theorems can simplify complex mathematical problems into elegant, closed-form solutions.