Aashu Singh
Backend EngineerAI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Senior Staff Software Engineer at Meta, MRS AI | LLM Post training
I lead the development of efficient MM-LLMs that extract semantic signals from multimodal content to improve relevance across Meta's recommendation and search stack. Our work focuses on building production-scale systems that process billions of content items for ranking, search and recommendation. - Designing lightweight multimodal models that operate at Meta's scale across billions of content items daily - Building systems that extract semantic and contextual signals from images, videos, and text for use in recommendation pipelines - Integrating multimodal representations into recommendation and search stack to improve content relevance metrics - Optimizing inference latency and computational efficiency for production deployment Previously, I built retrieval and ranking systems for Facebook Dynamic Ads, covering both ads and product catalog recommendation. Key work included: - Developing retrieval models for large-scale ad and product catalogs - Building ranking systems optimized for relevance and engagement - Designing user interest models that capture long-term preferences and short-term contextual signals across multiple timescales.
Georgia Institute of Technology
Master’s Degree, Computer Science
January 1, 2014 – January 1, 2016
Meta
Senior Staff Software Engineer (Machine Learning)
July 1, 2024 – Present
Meta
Staff Software Engineer
January 1, 2022 – January 1, 2024
Meta
Software Engineer
June 1, 2016 – March 1, 2022
Georgia Institute of Technology
Graduate Teaching Assistant - CS 4641 (Machine Learning)
January 1, 2016 – May 1, 2016
ADP
Software Engineering Intern
June 1, 2015 – August 1, 2015
Atlanta Metropolitan Area
Georgia Institute of Technology
Graduate Research Assistant
January 1, 2015 – February 1, 2016
Georgia Institute of Technology
Graduate Teaching Assistant - Machine Learning (CS 4641)
January 1, 2015 – May 1, 2015
TIBCO Software
Associate Member of Technical Staff
July 1, 2012 – July 1, 2014
Pune/Pimpri-Chinchwad Area
Detecting Social circles in Social Graph Data - Facebook
February 1, 2015 – May 1, 2015
Community detection on social graph data - Facebook. Performance improvements based on Louvain algorithm based on maximizing modularity. Tools: Python, Scikit-Learn
Context based Document Retrieval
December 1, 2014 – July 1, 2015
Enable Document Search based on semantics of the search query. Eg. Query string with "USA", also returns documents with America keyword. Performed Deep Learning in NLP to gain better representation of documents and queries as vectors and then calculated the similarity. Tools: Python, NLTK
Interactive Learning System (Project for CDC)
November 1, 2014 – Present
A hybrid recommendation system- content and collaborative type recommendation system. The ontology is used for query expansion and set of matching documents based on ranking algorithm is returned. User feedback is incorporated to personalize the recommendations. Tools: Python, ElasticSearch, MongoDB, StanfordNLP, NLTK
Improvement in RST Discourse Parsing
October 1, 2014 – December 1, 2014
Word Sense Disambiguation is performed over sentences/EDUs in text corpus using 'Wordnet' and ‘Adapted Lesk algorithm’ to identify the most suitable sense. Features were designed to capture sentences with similar senses. These features were then used for training SVM which led to improved performance of the discourse parser. The increase was observed across both identifying the correct EDU spans as well as relation between them.
Recommendation System Analysis
October 1, 2014 – December 1, 2014
Topic Modeling is performed on a reviews text of items from Amazon, using LDA. The latent factor dimension in LDA is kept equal to that used in collaborative filtering algorithm. The variations in the ratings of an item can be inferred from the topics obtained.
Performance Improvement of Ensemble Classifier
July 1, 2011 – July 1, 2012
Participated in a team based research project to study and improve the performance of the Random Forest Algorithm based on heuristics.
E-Commerce Simulation
September 1, 2010 – November 1, 2010
- It was a simulation to explain the functionality of online auction websites such as e-bay using Visual Basic and Oracle 10g. - Normalized database, concurrency control mechanism were some of the main features.
Cultural Fit Analysis
The candidate's extensive experience in research-oriented and large-scale production environments at Meta, coupled with academic research projects, suggests a strong fit for a culture that values innovation, data-driven decision-making, and tackling complex technical challenges. The diversity of projects, from NLP to graph data and e-commerce simulations, indicates adaptability and a broad interest in various technical domains. The progression within a single major company (Meta) for a significant period also suggests loyalty and ability to thrive in a structured environment.
Soft Skills & Operational Fit
The candidate's career progression at Meta from Software Engineer to Senior Staff Software Engineer (Machine Learning) demonstrates strong leadership, problem-solving, and project ownership skills. Their involvement in research publications indicates a proactive and innovative mindset. The teaching assistant roles suggest good communication and mentoring abilities. The focus on optimizing inference latency and computational efficiency aligns with operational excellence.