SMITA WADHWA
San Francisco,CA
Summary
Experienced software professional with over 16 years of expertise in 'designing and scaling platforms, infrastructure, and machine learning solutions' across 'Video processing, Image analysis, Advertising, and Big data domains'. Contributed to research efforts resulting in publications, patents, and industry recognition.
Skilled in tackling complex technical challenges, optimizing algorithms, and building scalable systems, with a strong focus on mentoring teams and fostering collaboration to drive organizational success.
Overview
18
18
years of professional experience
1
1
Certification
Work History
Staff Software Engineer
Twitter
10.2017 - Current
- Following Twitter's acquisition and a significant reduction of the Ads organization, I took on the leadership of multiple X Ads streams, including Ads Infrastructure, Ads Pacing, Ads Marketplace, Ads Serving, Ads Exchange, Ads Callback, and Ads Measurement and Reporting. Working closely with Elon Musk for X Ads, I prioritize a streamlined and efficient strategy to ensure uninterrupted service availability, meet revenue goals, and integrate new features seamlessly into the platform.
- As VideoAds Infrastructure Team tech lead : Developed corrective action plans and prevention strategies to improve the alerting system to minimize the on-call issues. Fostered relationships with videos and non-revenue customers for the video-ads teams projects and planning. Planned, scheduled, delegated quarterly tasks to the team. Devised , Implemented, configured and optimized ML models for Video Ads. Also worked on multiple components to scale up Video-Ads business of twitter and helped the team uplift the ctr by 18.5%.
- As Adserver Re-Architecture ideator/lead: Worked from scratch to a full-flegded design to reArch live Adserver . Brainstormed and got buy-ins from other leads and the leadership to decompose monolithic Ads backend to multiple independent services. Led one of the teams , collaborated with other engineers and leads, in different phases to develop and verify design changes. Managed team performance by training, mentoring, disciplining and motivating employees. Adhered to best design choices and standards to safely complete project TAO. Fostered relationships with internal and external teams.
- As Ads-Marketplace Individual Contributor : Re-Architected the ads-marketplace to be have rule-based engine which was config based instead of code. Took J-Easy open source library and adapted it to our requirements. Implemented this with one other peer and took it to scale of 0.5 million rps.
- As an Ads-science Individual Contributor : Drafted a plan for the new system to read 'ads-budget-pacing information' in 'asynchronous mode' irrespective of the request path. This design was first of its kind in Adserver system where revenue being burnt is completely dependent of budget being paced. Did fair share of thorough evaluation of the system, created a small prototype with 1% traffic using new caching scheme, got fair share of learnings and then planned, scheduled and delegated this to the team to productize it fully. Getting other leads on board to this new proposal was the first challenge of this project and rest is history.
- As an Ads Injection tech lead, currently working on creating a tech design and MVP for injecting ads in different timelines of twitter organic content. This library shall provide a unique platform to 'mix ads with organic content of twitter' while giving more visibility of ads being dropped to the business teams.
Technical Architect
inMobi
01.2012 - 10.2017
- As an architect/lead of Programmatic Market Place in inMobi SF(with team in SF and India), designed and resolved hard business problems while learning from a Google Mentor and collaborating with the team cross timezones(in India). Worked closely with business and product teams, and other stakeholders to understand the problems and mapped those problems to tech world while working on the probable solutions which required intense reading of prior research work. Implemented those problems with new ML based algorithms and Engineered architecture and infrastructure for @8 Billion queries/day
- Designed and implemented an automated data backed system which decides what publishers should be made available to each ad Network. This system increased inMobi’s overall revenue by 3% and saved hardware resources too.
- Wrote ML based Algorithm, Designed and Implemented for a dynamic pricing model of real time bidding setup. This project involved rigorous reading of prior research the solution was an advance implementation of 0/1 Knapsack algorithm . Improved business productivity for clients by increasing inmobi's revenue 6% by re-engineering and designing infrastructures. Collaborated and brainstormed with few industry pioneers to come up with the solution. Drafted, Planned and delegated the solution to the team.
- Designed and Implemented a modified Reinforcement learning Multi Armed Bandit algorithms(UCB and E-Greedy) for Dynamic Floor Management to decide reserved-price for different publisher/supply segments.
- Modified and implemented Data Clustering(based on Kmeans) for m-Commerce product which collects data from different e-commerce websites and clusters and tags them in different groups.
- As an architect of Video Ads in Bangalore, led the team of 10 people and worked on all phases of system development life cycle, from requirements analysis through system implementation for Video Ad Network like Ad Serving, Click Management, Billing and Reporting. Collaborated with clients to determine project specifications and scope. Wore multiple hats and did operations, QE, devOps etc tasks. Provided 2nd and 3rd level technical support and troubleshooting to internal and external clients.
Technical Lead Software Developer
Yahoo
01.2007 - 01.2012
- As an engineer of the 'one of the major components' of 'Platform that provides infrastructure as a service(that enables rapid development, testing, and release of software agility with stability), I worked on intelligent tier selection and placement for automated deployments on cloud system. Also Worked on elasticity of automated resource allocation in the cloud.
- As a tech lead , lead the team to design and implement Video Freshness Carousel(System to take feeds from multiple sources for fresh video results with the plugin to remove spam and boost fresh results for buzzy queries). The solution was based on Vespa topped with 'Real time Cache' for decreased latency. This required expertise on search engine relevance optimizations.
- As a tech lead of Video Platform team , worked on video spam detection problem on video.yahoo.com. Designed and Implemented the algorithm for video fingerprinting, video shot boundary detection and scaling the video fingerprints on index of 100+ Billion fingerprints as required by Yahoo! Video Content Analysis & Metadata Enrichment (Platform to analyze video content and enrich video metadata for video.yahoo.com and video.search.yahoo.com).
- As an individual contributor, worked on different computer vision projects like OCR , image recognition, frame detection, re-transcoding, rescaling of videos for internal video-streaming site(nodi for employees. This also motivated me to fix bugs in open source libraries.
- As a hack week project I worked on algorithm and developed for Video Delivery on Mobile which involved On the fly encoding with different codec support for huge range of mobile platforms(in 2008). This work got patent status too.
- As an internal project named as 'video preview generation' implemented a new algorithm for smart preview generation of the video using ffmpeg.
- As an individual contributor, worked on video language translator project using ffmpeg and sphinx.
- As an intern, worked on design and developed in c/c++ the algorithm for creating and matching video fingerprints after doing an intense research on current technologies , research solutions and current vendor solutions.
Education
Bachelor of Technology - Computer Science
Kurukshetra University
Kurukshetra08.2004
Master of Technology - Information Systems
NSIT
New Delhi, India06.2007
Skills
- Languages with professional experience : C, C, Java, Scala Perl, Python
- Big Data : Hadoop MR, Scalding, Spark, falcon/oozie
- Streaming/Caching: Kafka, Aerospike, mQueue
- Machine Learning : Unsupervised Learning(Clustering) , Supervised Learning(Classification and Regression) and Reinforcement Learning(Multi Armed Bandit Algos)
- Computer vision : opencv(OCR algos, image classification), ffmpeg(video transcoding, video fingerprinting, advertising )
- Others: svn, git, hudson, MySQL, postgres
- Systems Engineering with Design principles
- Conceptualizing development projects
- Redesigning scalable and live system architecture
Certification
- Optimizing callout in unified ad markets IEEE Big data 2016
- Distributed locality sensitivity hashing Mar 11, 2010 ACM
- Distributed Locality Sensitivity Hashing Jan 9, 2010 IEEE
- vADeo: video advertising system 2007, ACM and IEEE
Additional Information
- Video Encoding for Mobile Devices
http://patents.justia.com/patent/20100104004 - System, Method and Apparatus for Video Fingerprinting
Patent issuer and number us 20100061587 - Content Preview Generation
http://patents.justia.com/patent/20120050318
Accomplishments
- One of the architects for re-Architecting(monolith -> microservice) Twitter's live adserver acknowledged in industry.
- One of the 12 Inspiring Architects in Women2 in 2017.
- Nominated for Worldwide TR35 2010 young Innovators.
- Published few patents and publications on my work done in industry.
- Contributor for bugs and features of Open Source Apache Mahout Project.
- Winner of 3 consecutive Y! Hack Days for 3 projects. Awarded Yahoo Quarterly ‘U Rock’ for Best Idea, Innovation and Practice in 2008
- University Topper during Masters of Technology.
- All India GATE 2005 score 96.77 in CS All India second topper in science in tenth CBSE boards with 99%(science).
Timeline
Staff Software Engineer
Twitter
10.2017 - Current
Technical Architect
inMobi
01.2012 - 10.2017
Technical Lead Software Developer
Yahoo
01.2007 - 01.2012
Bachelor of Technology - Computer Science
Kurukshetra University
Master of Technology - Information Systems
NSIT
- Optimizing callout in unified ad markets IEEE Big data 2016
- Distributed locality sensitivity hashing Mar 11, 2010 ACM
- Distributed Locality Sensitivity Hashing Jan 9, 2010 IEEE
- vADeo: video advertising system 2007, ACM and IEEE
Similar Profiles
EDGAR IBARRA, JR.EDGAR IBARRA, JR.
Business Operations Specialist at TwitterBusiness Operations Specialist at Twitter
BRITTAIN LADDBRITTAIN LADD
Business Advisor at TwitterBusiness Advisor at Twitter
ALEKSANDR SHULMANALEKSANDR SHULMAN
Software Engineer at TwitterSoftware Engineer at Twitter