Farhan Hyder
Software Engineer at Meta with 6+ years of experience across full-stack development, cloud infrastructure, and LLM-powered systems. From founding engineer on a fintech alternative funds platform to privacy infra at Meta, I've shipped production-grade ML systems, built and owned end-to-end backend pipelines, and driven measurable performance improvements across complex distributed systems. Currently pursuing an M.S. in Computer Science (ML specialization) at Georgia Tech part-time.
Experience
Meta, Bellevue, WA
- Own privacy policy detection within Diff Time Signals, a runtime analysis system at Meta that evaluates every engineering code change before it lands; reduced average detection runtime from 90+ minutes to 51 minutes, accelerating the diff feedback loop for engineers.
- Designed and productionized a hybrid LLM and heuristic classification system that identifies third-party vendor model responses, achieving 96%+ precision and recall and eliminating 127+ engineering hours per week of manual review.
- Built reusable automated ground-truth evaluation infrastructure for the team's classifiers and agent systems, replacing manual performance tracking with structured metric pipelines.
- Built performance analytics dashboards giving engineers visibility into LLM behavior and system quality metrics.
- Drove cross-system domain migration from legacy event pipelines to a zone-based architecture, lifting functional parity from 77% to 99.5% by resolving long-tail edge cases.
Accenture, New York, NY
- Founding engineer of BNY Pershing's Alternative Funds Platform (now Alts Bridge); contributed across design, prototyping, and stakeholder iteration through production launch, delivering advisor access to alternatives from managers including KKR, Apollo, and Blue Owl Capital.
- Designed and shipped the advisor-facing order entry workflow: a multi-page dynamic form with conditional branching, draft persistence, and approval-routed signing.
- Built and owned the fund portfolio management pipeline: scheduled ingestion of fund data and documents from external sources, automated validation, and pre-processing gates feeding order entry.
- Built advisor-facing AI features for the platform: integrated a RAG-based agent for fund summaries, updates, and Q&A; designed per-advisor context persistence so each advisor's history and interaction state carried across sessions.
- Drove integration between Alts Bridge and BNY's Wove wealth management platform, validating each new capability against legacy system contracts and data flows.
- Played a key role in designing, developing, and implementing functional components that enable clients to manage cross-cloud applications (AWS, Azure, GCP) directly from the ACF platform.
- Designed and implemented a reusable component using AWS SNS and Lambda as a workflow queue system, driving a major internal architectural shift from monolithic to decoupled. This improved system maintainability and made it easier to add new features, while also allowing other teams to use the component.
- Managed the transition of project leadership, maintaining continuity and excellence in project delivery, and ensuring a smooth handover from the China dev team to new team members in India.
- Conducted extensive training and mentorship, onboarding more than ten new team members on feature-related tasks and offering in-depth guidance on ACF cross-cloud platform.
- Orchestrated collaboration across diverse international cross-functional teams of over 45 members, ensuring smooth operations and development processes among teams from Argentina, China, Czech Republic, India, Ireland, Mexico, Philippines, and the USA.
- Played a pivotal role in the full-stack development and release of over 35 critical features, leveraging Angular for front-end, NodeJS for back-end, and AWS for scalable cloud deployment.
- Enhanced the GCP sandbox project auto-decommission feature, achieving a 97%+ success rate in identifying and removing redundant projects, significantly optimizing resource utilization.
- Implemented Rebar Messaging Service within ACF cross-cloud platform, enabling more efficient inter-service communication.
- Created internal and external APIs with NodeJs and AWS Lambda function.
- Refined and optimized CI/CD pipelines within Azure DevOps and AWS CodePipeline.
- Coordinated with multiple analytics teams to develop and integrate report pages and dashboards in ACF cross-cloud platform, significantly improving user access to critical data insights.
- Developed and maintained automated scheduled jobs using AWS CloudWatch, and Lambda as part of provisioning projects, retriggering failed jobs, clean ups, and audits.
Western Union, Edison, NJ
- Designed, developed, and maintained Java Servlets for web services, monitored traffic patterns and implemented automated solutions to improve site reliability and system availability.
- Collaborated with external teams to plan and perform the migration from legacy (on-prem IBM DB2) to cloud (AWS RDS).
City University of New York, Long Island City, NY
- Mentored college students in Advanced C/C++ Programming, Object-Oriented Programming with Java, College Algebra, and Trigonometry.
- Prepared lecture notes, quizzes, programming assignments, and mock tests.
- Worked with other Academic Peer Instruction Tutors to improve tutoring skills and research on different effective education models.
Skills
| Languages | Java, Python, JavaScript, TypeScript, SQL, C++, Hack |
| Back-end | Spring Boot, Node.js, Express, GraphQL, PySpark, PostgreSQL, MongoDB, Neon, Supabase |
| Front-end | Angular, ReactJS, HTML, CSS, NextJS |
| Cloud/Infra | AWS (DynamoDB, Lambda, S3, CloudWatch, SNS, Step Functions, CodePipeline), Azure (DevOps, Functions, Blob Storage), GCP (Cloud Functions, Firestore, Authentication) |
| ML/AI/Tooling | LLM evaluation, classification pipelines, agent observability, prompt engineering (Claude Code, Cursor) |
| Tools | Agile, TDD, Git, SVN, Jira, Splunk, Dynatrace, Numpy |
- English, Bengali - fluent
- Hindi, Urdu - converse
Projects
- Stock Trading Calculator | Web Application |
- Chitty Chat | Web Application |
- Pooled House | Web Application |
- Pizza Hub | Web Application |
- Shared Music Player | Console Application |
- Merit Evaluation of Different CPU Scheduling Algorithms | Console Application |
- Tip Calculator | Android app |
A website that lets a user to save their stock information and perform different calculations to ease their stock trading. Full-stack angular development has been used with Google Cloud Services. Github is used for version control, and Bootstrap for partial front-end UI design.
Stock Trading Calculator Demos:A chat room messaging service that uses machine learning to detect and analyze the tone of a message and convert that into an emoji. This service was built while working in an agile software development team for a college project. I acted as a Software Architect, also responsible for continuous deployment, back-end development, and tone-analyzer middleware API creation.
Chitty-chat Demos:A web application that collects, compares and gives multiple visualizations of tip data from NYC service industry. The website was created as a part of a project in the Web Site Design class at CCNY. I acted as a scrum master for this project. I was responsible for managing a team of five to achieve weekly goals, implementing map-view using MapBox, and handling backend with AWS AppSync.
Pooled House Demos:A group project for Software Engineering course at CCNY. A Mid-Manhattan pizza ordering website, that takes orders from the customers, then sends the order to the selected store. The store-management-site is designed for three types of users- manager, delivery guy, and cook. JavaScript and Node.js are used for the web application.
I was fully responsible for the Google Maps API, Searching Nearest Stores, and Graphics. Shared responsible for System design and Data-flow. Following are the samples of some of my work in this project.
A real-time shared music player between two users. Both users have the same permissions in terms of changing the songs. Pthreads has been used to prevent any racing conditions, TCP Sockets has been used for Inter-Process Communication.
Shared Music Player Demos:A project to evaluate the merits of FCFS, SJN, and Priority scheduling with and w/out Round-Robin by taking the Average turnaround, Average wait, Longest wait, and Total run time of each. C, C++ has been used to construct the scheduling algorithms, and Python has been used to plot the graphs.
Different CPU Scheduling Algorithms' Merit Evaluation Demos:The app asks from the user to input the total amount of bill and number of people they would like to share the amount with. It also provides quick tip amounts 15%, 20% and 25% to the help the user with a quick summary.
I have used the software Android Studio to make this app, used JAVA as the main language and XML for the graphical interface.
Education
Georgia Institute of Technology
- Machine Learning for Trading (CS 7646)
- Global Entrepreneurship (CS 8803)
- Intro to Cognitive Science (CS 6795)
- Artificial Intelligence (CS 6601)
- Game AI (CS 7692)
The City College of New York
GPA: 3.1
- Artificial Intelligence
- Intro to Data Science
- Image Processing
- Web Security
- Website and Web Applications
- Computer Language Paradigms
- Operating Systems
- Software Engineering
- Software Design Lab
- Theoretical Computer Science
- Numerical Issues in Scientific Programming
- Database Systems
- Fundamentals of Computer System
- Algorithms
- Discrete Math Structures
- Data Structures
- Probability and Statistics for Computer Science
CUNY LaGuardia Community College
GPA: 3.8
- Computer Organization and Assembly Language
- Object Oriented Programming with Java
Certifications
Links
I like to experiment with my creative abilities from time to time. Here are a few samples...
- made for a group of senior Mechanical Engineering students at CCNY