Building software and models that forms a crucial part of Liminal's ultrasound-based solution for advanced inspection, process control, and process optimization for battery manufacturers. Battery manufacturers use Liminal's solution to improve battery performance, reliability, and safety while reducing overall battery costs.

Took some time off to recover from burnout and take care of my mental health.My desire to learn and build never really went away, so I ended up developing an interest in climate change, taking Terra.do's flagship course on climate change, and built the Obsidian.md plugin for Pocket, which is used by over 6K people.I eventually felt limited by what I could do alone, so I'm looking to join a team that I can learn from and build ambitious things with again.

Allocated 6-8 months to spend on incubating ideas, with the aim of monetizing one and building a company out of it.Learned about various customer problems, and looked for software-based solutions. Eventually focused on the "tools for thought" space (i.e. products like Notion and Obsidian.md, among others). Conducted market research and user research. Designed and built prototypes using Google Cloud Platform, Firebase, Typescript and Material-UI. Put together marketing funnels for lead generation and user acquisition.Decided to stop our efforts after we were unable to meet our goals for traction and conviction after the allocated time was up, and open-sourced our efforts.

Acted as the voice of the customer team in CAR's portfolio of research programs, ensuring that research efforts were aligned with customer teams' medium and long-term needs. This was done through various stages of research work, from customer problem formulation and research program proposal to setting up operational scaffolding for healthy collaborations with customer teams.Acted as the voice of CAR with customer teams, ensuring that stakeholders were aligned, that research programs had measurable and meaningful goals and that commitments (e.g. engineering resources) were clear. Communicated CAR and its impact and ongoing work to other stakeholders (e.g. Cortex senior leadership).Worked with CAR leadership team (PMs and EMs primarily) to shift from team-based efforts to research program-based efforts, so that CAR could take a customer-problem-focused approach and leverage the right research skills from across the pillar.Collaborated with other PMs and with EMs for quarterly planning, i.e. understand research programs' progress and impact, then prioritize and resource for upcoming quarters.Towards the end of my tenure as a PM on CAR, I realized that I really wanted to try something new, after having been at Twitter for eight and a half years.

Transitioned from an engineering management role in the User Growth organization to an individual contributor role, in Cortex, which is the central machine learning infrastructure and applied research team within Twitter.Worked on machine learning infrastructure and machine learning-powered products across Twitter.Note: I spent May 2015 to Jan 2017 as a Senior Software Engineer in this role, and was promoted to Staff Software Engineer in Feb 2017.HighlightsML product:- Partnered with the platform safety team to apply machine learning to directing Twitter's enforcement resources to the most egregious terms-of-service violations, drastically reducing the number of unaddressed terms-of-service violations (>5x decrease).ML platform:- Built the capability to handle text, images and other "dense" features in Twitter's deep learning infrastructure, which were subsequently used extensively in Twitter's efforts to apply machine learning to platform safety, amongst other uses.- Contributed to many Cortex production ML systems by building data pipelines and data sampling/hydration layers, including systems for ad safety and platform safety.Applied ML research:- Conducted applied research on natural language models based on tweet text, which allowed Search Quality to surface more relevant tweets faster, regardless of their popularity.Technical leadership:- Formulated and led a working group to improve the ease-of-use of Cortex's offerings by conducting user interviews with customer teams. Cortex teams then used our findings to inform their efforts in improving ease-of-use of the Twitter machine learning platform.- Developed a course to introduce a broad audience to applying machine learning to solve business problems, scaled it to 14 instructors and delivered it to over 1300 attendees.- Led multiple cross-organizational projects to migrate from deprecated machine learning services to modern and performant ones.

Transitioned from the TLM role to focus solely on engineering management. Started the Contacts Graph team with one engineer. Grew the team to twelve engineers (including two TLs) over the course of a year, through recruiting and hiring both internally and externally. The team owned and built out two key initiatives: the Contacts Graph and the Instant Timeline.The Contacts Graph initiative was to modernize Twitter’s client code, APIs and infrastructure for address book data and provide tools for Twitter users to manage their address book data (https://blog.twitter.com/en_us/a/2015/a-new-dashboard-to-help-you-monitor-and-manage-your-twitter-account). The address book infrastructure we built also powered the friend-finding functionality in the Digits SDK (https://blog.twitter.com/developer/en_us/a/2014/a-better-way-to-sign-in-with-digits).The Instant Timelines initiative was an early attempt at solving the new user onboarding problem at Twitter by allowing new users to have a ranked timeline of tweets, without actually following any accounts (https://bits.blogs.nytimes.com/2015/02/02/twitter-displays-its-value-with-instant-timeline-for-new-users/).I also provided technical leadership when needed and fostered a healthy and empowering team culture through mentorship, regular retros/demos and through performance management.Our team reached a juncture where both initiatives were complete. I decided to transition back to a individual contributor role and have the team members transition to already-existing initiatives within the User Growth organization so that the organization as a whole could continue to focus on a small number of high-impact projects.

Continued work in service of driving the engagement and growth of Twitter’s user base. Transitioned to a Tech Lead role. Led full-stack feature projects involving multiple engineers. Hosted and mentored software engineering interns, and participated in recruiting and hiring efforts.Eventually transitioned to a formal Tech Lead Manager (TLM) role with three direct reports, to provide technical leadership and people management and execute on foundational work for the address book features and infrastructure at Twitter.

- Built features to drive the growth and engagement of Twitter's user base.- Designed, built and maintained scalable product infrastructure for recommendations, address book data, contact import, invitations and other use cases related to relevance and user growth.- Developed and refined machine learning algorithms for engaging new users with relevant content.- Conducted data science work to inform product roadmaps and evaluate project success.- Designed metrics and built them into analytics pipelines for monitoring user growth behavior.This role involved leveraging a wide variety of technologies, including Hadoop, Pig and Java (for recommendations and data science work), Scala/Finagle, Clojure/Lamina/Aleph and Mesos (for scalable product infrastructure) and Objective-C, Javascript and Ruby (for frontend work).

Joined Summify as the second employee. Worked on the entire Summify product, which delivered stories from Twitter and Facebook directly to our users' inboxes. Acquired by Twitter in January 2012. I then started at Twitter as a Software Engineer on the User Growth team.This role focused on full-stack development. The stack consisted of Python, Django and Tornado, with MySQL, Redis and Mongo used as data stores.

Conducted research at the intersection of natural language processing (NLP) and product recommendations. Built and tested prototypes of various NLP algorithms. Integrated performant solutions into Optemo's production systems.This work was supported through a MITACS Accelerate scholarship.Used Python, Numpy, NLTK, SQLite and MongoDB for research prototyping work. Integrated performant solutions into a Ruby-on-Rails-based production system.

Worked on the platform team of BusinessObjects, a business intelligence product. This role focused on high-performance cross-platform (Windows and a variety of Unices) software written in C/C++.

Worked on an early embedded systems engineering effort to bring video playing capability to 3G mobile phones. Worked on bug squashing and automated testing in a messy C/C++ codebase.Built a testing automation system that allowed users to programmatically build and run testcases on 3G mobile phones, as an alternative to the widespread process of testing using manual phone keypad presses.

Worked on a distributed traffic simulator used in the development of one of the first deep packet inspection routers. Developed FreeBSD kernel extensions in C. Developed libraries in Tcl and C/C++ for VLAN configuration and stateful P2P network traffic generation.

Developed software for Linux-based zero-configuration office network servers aimed at SMBs. Worked on both low-level OS-level code (C/C++) and higher-level orchestration layers (Python, Perl).

Wrote documentation and technical manuals for software used by loan officers to make decisions regarding loans for purchasing farming equipment.