PhD candidate researching Natural Language Processing. I am mostly excited about building and understanding systems that reason to perform tasks that are helpful and complex, such as multi-hop question answering and multi-modal reasoning. My most recent works focus on improving performance, efficiency, and robustness of retrieval-augmented large language models.
In my industry position, I am a founding team member and research lead at Ask-AI, where we are developing a generative AI sidekick that reasons over enterprise data. I was also fortunate to intern at Microsoft and AI2 during my studies.
About
PhD candidate researching Natural Language Processing. I am mostly excited about building and understanding systems that reason to perform tasks that are helpful and complex, such as multi-hop question answering and multi-modal reasoning. My most recent works focus on improving performance, efficiency, and robustness of retrieval-augmented large language models.
In my industry position, I am a founding team member and research lead at Ask-AI, where we are developing a generative AI sidekick that reasons over enterprise data. I was also fortunate to intern at Microsoft and AI2 during my studies.
Publications
News
Ask-AI raises 11M$ in A Round funding
January
Attending EMNLP in Singapore
December
Published a new paper on how we can make retrieval-augmented language models robust to irrelevant context (Update: accepted to ICLR)
October
Part of the effort to release the RippleEdits benchmark for evaluating if knowledge editing methods can capture complex “ripple effects” (Update: accepted to TACL)
July
Attending ACL in Toronto
July
Published our paper on how meta-reasoning over multiple Chains-of-Thought can be a great way to improve current systems for complex open-domain question answering (Update: accepted to EMNLP)
April
Started my PhD! My research will be focused on the ability of tool- and retrieval-augmented Large Language Models to reason
November
Part of the effort to release QAMPARI, a benchmark for evaluating the ability of current open-domain question answering systems to answer questions with many answers
May
Part of the effort to release the SCROLLS benchmark to evaluate the reasoning skills of Language Models over long tests (Update: accepted to EMNLP)
January
Joined the Ask-AI founding team raising a 9M$ seed-round to build a Gen-AI sidekick that reasons on text-heavy enterprise data
December
Graduated my MSc (Magna Cum Laude)
October
Published our work on using gamification to challenge commonsense reasoning of LLMs (Update: accepted as an oral presentation to NeurIPS)
July
Published my MSc thesis on how we can use tabular data to automatically generate training examples that can endow language models with reasoning skills (Update: accepted to ACL)
July
Published our work introducing the MultiModalQA dataset for evaluating multi-modal question answering (Update: accepted to ICLR)
April
Started an internship with the AI2 gamification team, researching how simulating games can be used for data collection
October
Started my MSc thesis working advised Prof. Jonathan Berant on enhancing the reasoning skills of Language models
August
2023
2022
2021
2020
2024
News
2021
2022
2020
2023
2024
January
Ask-AI raises 11M$ in A Round funding
December
Attending EMNLP in Singapore
October
Published a new paper on how we can make retrieval-augmented language models robust to irrelevant context (Update: accepted to ICLR)
July
Part of the effort to release the RippleEdits benchmark for evaluating if knowledge editing methods can capture complex “ripple effects” (Update: accepted to TACL)
July
Attending ACL in Toronto
April
Published our paper on how meta-reasoning over multiple Chains-of-Thought can be a great way to improve current systems for complex open-domain question answering (Update: accepted to EMNLP)
November
Started my PhD! My research will be focused on the ability of tool- and retrieval-augmented Large Language Models to reason
May
Part of the effort to release QAMPARI, a benchmark for evaluating the ability of current open-domain question answering systems to answer questions with many answers
January
Part of the effort to release the SCROLLS benchmark to evaluate the reasoning skills of Language Models over long tests (Update: accepted to EMNLP)
December
Joined the Ask-AI founding team raising a 9M$ seed-round to build a Gen-AI sidekick that reasons on text-heavy enterprise data
October
Graduated my MSc (Magna Cum Laude)
July
Published our work on using gamification to challenge commonsense reasoning of LLMs (Update: accepted as an oral presentation to NeurIPS)
July
Published my MSc thesis on how we can use tabular data to automatically generate training examples that can endow language models with reasoning skills (Update: accepted to ACL)
April
Published our work introducing the MultiModalQA dataset for evaluating multi-modal question answering (Update: accepted to ICLR)
October
Started an internship with the AI2 gamification team, researching how simulating games can be used for data collection
August
Started my MSc thesis working advised Prof. Jonathan Berant on enhancing the reasoning skills of Language models