Anirudh Seth

Co-founded a venture-backed startup. Stealth for now — happy to share more in conversation.

Lead the engineering organization responsible for Dropbox Dash, AI, and Experiences — the AI-powered platform at the center of Dropbox's product strategy. Founded and launched Dropbox Dash. Joined Dropbox to lead the incubation team chartered with building Dropbox's next-generation AI products. Took Dash from concept through architecture, build, and launch — including the AI/ML platform, search and retrieval infrastructure, connectors platform, and knowledge graph. Integrated the Nira acquisition and launched Protect & Control. Led the end-to-end engineering integration of Nira into Dropbox, rebranded as Protect & Control and shipped as part of Dash for Business. This was the fastest acquisition-to-launch in Dropbox's history. Brought AI to core Dropbox. Led the engineering effort to integrate AI capabilities into Dropbox's core file sync and share product. Key technology areas: enterprise knowledge retrieval, permission-aware search, knowledge graphs, RAG architectures, connector infrastructure across 30+ SaaS integrations, LLM orchestration with DSPy.

Led engineering and applied science teams to build Magic Button — a global Demand Side Platform (DSP) for mobile and PC game developers. Built the team, infrastructure, and ML models from the ground up, including a data platform processing over 1 petabyte daily. Crafted the vision and strategy to reach $1B in ad revenue by 2026 and secured executive sponsorship (SVPs/VPs across Amazon Ads and Amazon Games) for a $6.9M budget increase. Enabled mobile install ad campaigns for game developers leveraging cloud technologies and machine learning. Collaborated cross-functionally with Legal, Privacy, Policy, Business Development, and Account Management to define SOPs for data handling, money flow, customer incentives, and account management.

Established and led a team of 12 engineers in delivery of a data driven Reverse logistics routing system that processed disposition of 1 billion items annually. The system determined the best revenue and sustainability path by selecting the most "optimal" second life for the products (I.E. Donation of goods during Covid). Authored a three-year vision for the product, generated executive consensus, and directed team in construction of a routing system comprised of multiple micro-services. • Increased net recoveries by $10.5 million within the first five months of launch by generating optimal recovery channels. • Facilitated Amazon's sustainability goal by allowing donations (rather than destruction) of 1.2 million items worth $75 million during Covid (https://good360.org/press-releases/good360-distributes-more-than-125-million-in-needed-goods-during-covid-response) by supplying an effective technology solution. • Attained a team attrition rate of only 8%, compared to the department average of 40%.

Directed a staff of five (two Principal Engineers and three Lead Engineers) that supported strategic business investments in big data analytics and its application in TV Everywhere and CVML Retail Analytics. Drove implementation of a content and ad recommendation system for the TV Everywhere solution. Engineered a system for the Retail Analytics system to provide insights, predictions, and recommendations based on data produced from sources like camera, IOT devices, Bluetooth, Wi-Fi, and point of sale. • Enabled implementation of a new Business Unit (Ittiaminsite) by defining a product and strategy for Computer Vision and Machine Learning initiatives supporting Retail applications.

Led three teams of eight internal developers and over 20 outsourced developers supporting the Android client and the core technology for Tribeway (an interactive sharing app - pictures and chat) that leveraged propriety image transmission technologies and state-of-the-art acoustics package. Drove delivery of the Tribeway app from conceptualization to launch. Researched mobile video communication/collaboration market and made the internal sales pitch including the solution, SWOT analysis, and budgeting. Architected Android client (application layer, signaling, media and user management modules) and the client media system architecture that handled video based picture sharing and N-way Voice Bridge. Evaluated UI implementation RFPs, oversaw vendor delivery, and the User Acquisition and PR strategy. Spearheaded the Android Playstore release including all peripherals (website, legal terms, FAQs, payment accounts, privacy clauses, ASO). • Achieved a 100K WAU and Top 20 ranking in Google Playstore Apps in the first quarter of releasing the Tribeway app. • Reduced bandwidth requirements 2000% by designing an image transmission technology that yielded the highest quality, ensured minimum latency, and at lowest possible bandwidth, including a scheme for dynamic bandwidth adaptation. (Patent applied)

Video Communications Systems , Video Communication and Networking • Led three teams comprising of 18 engineers of varied years of experience and delivered the Video Communication Media System for all customer/roadmap projects • Designed and delivered Low latency Video streaming solution • Multiple Design Copyrights : Built Intellectual Property which was copyrighted by our customers like Multi-processor AV syncing, Active speaker highlighting using audio beam forming, Dynamic Intra refresh in error prone networks • Technical Proposals and Trade Shows (CES/MWC): Prepared technical proposals for potential customers and defined features to showcase in trade shows. • Implemented processes which targeted increased efficiency, maintainability of code, increased knowledge and growth of the team members, which eliminated the 30% overrun and helped in increase employee satisfaction and retention. • Reduced Delivery Cost by 25%: Used the Pareto principle to identify the top 20% items contributing to 80% of the cost, Platform porting being one of them. Redesigned the architecture to be platform agnostic. This halved the porting time.

HD Multi Codec Decoder RTL (H.264, Mpeg2, Mpeg 4 , VC1, Jpeg, RM8/9, WMV, H.263, AVS) • Contributed towards key features of a multi-standard hardwired video decoder (1 million+ gate design) • Led teams comprising of people with different levels of experience to implement key features such as API definition and Firmware development for the Video System. • Innovated a new concept in multi-million gate design by using SIMD (Single Instruction Multiple Data) and TDM (Time Division Multiplexing) with 2 different clocks. This design won accolades in the larger VLSI industry.