Mayank Daswani

Software Engineer

About Me

My name is Mayank Daswani and I’m a software engineer currently working in cyber security at QuintessenceLabs. In my spare time, I try to keep up with the literature on AI Safety and deep reinforcement learning. In a past life, I’ve completed a PhD in Reinforcement Learning with Marcus Hutter. You can find a list of papers I’ve published on my Google Scholar profile. If you want to get in touch with me, feel free to do so via LinkedIn or fill in my email via the template <firstname><lastname>@gmail.com.

Experience

QuintessenceLabs

Software Developer, Technical Lead

August 2015 - present

Founded in 2008 in Canberra, Australia, QuintessenceLabs offers the strongest security foundation for your data. This includes centralized enterprise key and policy management, a high-speed true random number generator, an integrated hardware security module, and highly secure encryption for data in uncontrolled environments.

I work on the qCrypt team at Qlabs. qCrypt is an Enterprise Key and Policy Management solution and our flagship product. Backing the software is a hardware platform that includes the world’s fastest (1Gbit/s) quantum random number generator (QRNG), and a third-party Hardware Security Module (HSM).

I’ve worked on or have been the technical lead for a wide-range of product features and deliverables. Some notable projects include helping design and implement a multi-master replication system, third-party HSM integration, and improvements to our Backup/Restore functionality.

Education

Australian National University

PhD Computer Science

2011 - 2015

The Australian National University, founded in 1946 through an Act of Parliament, is consistently ranked in the top two universities in Australia and in the top 30 worldwide. Known for its research excellence, it is situated in Canberra, the quiet leafy capital of Australia.

I completed a PhD under the supervision of Marcus Hutter and Peter Sunehag. My doctoral thesis is titled Generic Reinforcement Learning beyond Small MDPs.

The broad theme of my work was to find practical ways of solving large partially observable environments using a technique known as Feature Reinforcement Learning (FRL), a framework within which an agent can automatically reduce a complex environment to a more tractable Markov Decision Process (MDP) by finding a map which aggregates similar histories into the states of an MDP. My focus was on empirical work targeted at practitioners in the field of general reinforcement learning, with theoretical results wherever necessary.

Australian National University

Bachelor of Computer Science (Honours), University Medal

2007 - 2010

I finished an undergraduate degree at the ANU in Computer Science doing the prestigious Bachelor of Computer Science degree (equivalent to the current degree Bachelor of Advanced Computing (R & D)). As part of this degree, I was able to craft a largely custom course through the Computer Science and Mathematics courses at the ANU.

My degree had an built-in honours year, I submitted my honours thesis in the field of reinforcement learning titled “An Empirical Evaluation of ΦMDP Agents”. For my performance through my degree, I received the prestigious University Medal .

Projects

Where Should I Live?

whereshouldilive.space

Govhack grew from a 2 city event in 2012, an 8 city event in 2013 and a national 11 city event with over 1300 participants and observers in 2014. With 2016 culminating across 40 locations and over 3000 competitors and observers in two nations.

Govhack is a 40-hour hackathon held across Australia and New Zealand, with a goal of doing cool new things with government data. Where Should I Live was my 4-person team’s entry into Govhack 2017. It tries to solve the problem of deciding which suburb to live in focusing on Canberra, Australia. The user can select the importance of different factors to them, generating a customised heatmap indicating the livability of different suburbs. It is built in React, using Leaflet JS to present the heatmap. We collated data from 11 different data sources to form our livability score.

Getting a new business name is a pain, because you need to do three different searches: check the Australian Business Register that the business name is free, check the IP Australia trademark database that your name hasn't been trademarked in your industry, and check for appropriate domain names related to your name.

Biznamr is another Govhack project submitted with a different team of 3 (Evronos) in 2016. It presents a single interface for doing searches of multiple databases - the Australian Business Register, the IP Australia trademark database and a direct URL search. The data from IP Australia is live and comes directly from their API. It ranks the trademark results based on the keywords you provide to determine the most likey category. It does this using a feedforward neural network. Biznamr won us two prizes at Govhack 2016, first place in the category “Fresh Data Hack” and second place in “Best Entrepreneurial Hack”.

A new kind of hedge fund built by a network of data scientists.

When Numerai first started in 2016, I played with several neural network models mainly based on Highway Networks and was able to do well for a few weeks in the top 100. My profile above shows I was able to accumulate about USD $3500 during this period of time.

A Little More About Me

Outside my interests in machine learning, software engineering and reinforcement learning some of my other interests and hobbies are:

  • Indoor Bouldering
  • Running
  • Pentesting