Tejas Khairnar

AI Researcher

Amsterdam, North Holland, Netherlands6 yrs 9 mos experience

Most Likely To Switch

Key Highlights

Expertise in quantitative research and machine learning.
Developed advanced models for leading tech companies.
Represented India in the International Collegiate Programming Contest.

Stackforce AI infers this person is a Fintech and SaaS expert with a focus on quantitative research and machine learning.

Contact

Skills

Core Skills

Quantitative ResearchMachine LearningDeep LearningNatural Language Processing (nlp)Information RetrievalLeadershipComputer Vision

Other Skills

Apache SparkC (Programming Language)C++CryptocurrencyData Structures and algorithmsHigh-Frequency TradingLogical ApproachManagementMathematicsMySQLPyTorchPythonPython (Programming Language)SQLShell Scripting

About

I am a Quantitative Researcher at Portofino Technologies, where I apply my mathematical expertise, coding proficiency, and fascination with data-driven solutions to develop and optimize quantitative models and strategies for various financial markets. I have a strong background in machine learning, deep learning, and C++, having worked on several projects involving generative AI, chatbots, and search engines at Sprinklr and Amazon, where I achieved impressive results and recognition. My passion for mathematics and coding began in my childhood, when I specialized in Olympiad mathematics, particularly in enumerative combinatorics, and honed my problem-solving abilities by actively engaging in competitive programming. I represented India in the prestigious International Collegiate Programming Contest (ICPC), which was a defining moment in my academic journey. I graduated from Indian Institute of Technology, Guwahati, with a Bachelor of Technology degree in Computer Science in 2022, and completed multiple certifications and publications in the fields of deep learning and natural language processing. I am always eager to learn new skills and technologies, and to leverage them to drive meaningful outcomes. I am looking for opportunities that allow me to collaborate with other passionate and talented professionals, and to tackle complex challenges in the domains of quantitative research, machine learning, and large-scale systems.

Experience

Quantbox research

Quantitative Researcher - Crypto

May 2024 – Present · 1 yr 10 mos · Amsterdam, North Holland, Netherlands · On-site

Portofino technologies

Quantitative Researcher - ML Alpha

Mar 2023 – Mar 2024 · 1 yr · Amsterdam, Netherlands / Zug, Switzerland · On-site

Signals ExtractionMathematicsQuantitative ResearchCryptocurrencyPythonC+++4

Sprinklr

Product Engineer - ML/AI Team

Aug 2022 – Feb 2023 · 6 mos · Gurugram, Haryana, India

Worked on Generative AI driven end-to-end ML pipelines for conversations, catering over 1000 clients.
Developed chat-bots having capabilities of smart reply, auto-complete, grammar correction with less than 200ms latency.
Engineered various dataset sampling techniques using clustering increasing variance in the dataset by 10 times

Deep LearningNatural Language Processing (NLP)Data Structures and algorithms

Amazon

Applied Scientist - Search Team

May 2022 – Aug 2022 · 3 mos · Bengaluru, Karnataka, India

Developed a SBERT based model to retrieve the relevant titles for a given user query in Prime Video.
Performed detailed analysis of 100+ Gigabytes of data using PySpark and SQL and carried out rigorous ablation studies.
Transformed the actual Prime Video titles to generate proxy user queries to get the dataset with relevant query-titles.
Worked got accepted in ACM Web Conference 23 where 2.5x recall@16 was achieved during the evaluation of the model.

Python (Programming Language)Deep LearningInformation RetrievalSQLPyTorchNatural Language Processing (NLP)+2

Sprinklr

Product Engineering Intern - ML/AI Team

May 2021 – Jul 2021 · 2 mos · Gurugram, Haryana, India

Built a Computer vision based NSFW content filter to improve the recall and reduce the bias in the existing model.
Compiled a dataset with over 150 categories of images by web scrapping and removed noise using statistical analysis.
Transformed the EfficientNet baseline model to get a smaller model EfficientNet b(-4) which was used as the student
model during Knowledge Distillation from EfficientNet b0 as the teacher model.
Evaluation of the student model reflected a 4x smaller memory footprint and a 2x reduction in the inference time.

Python (Programming Language)PyTorchWeb ScrapingComputer Vision