a little about me,

about

hello! （＾ω＾）

• i'm tushar, 22, a software engineer and a researcher.
• my research interests are nlp, llms, information retrieval, and dl.
• most of my learnings are from either hands-on projects that i build in hackathons or from interesting courses taught at my university.
• tags: problem solver, clear comms, good vibes only.

tech stack

skills

• programming languages: python, c, c++, swift, sql
• databases: supabase, mongodb, neo4j, chromadb, coredata, mysql
• expertise area: dsa, oops, ml, dl, nlp, llms, transformers, rag, cv, mlops, information retrieval
• tools & tech: tensorflow, pytorch, keras, mlc engine, streamlit, langchain, litellm, huggingface, faiss, aws, gcp, azure, scikit-learn, docker, redis, nats jetstream, inngest, opentelemetry, grafana, fastapi, flask, git, selenium, postman

experience

profile

applied ai engineer

spotgov | lisbon, remote
oct 2025 - present

• led procurement ingestion workers for spain, brazil and eu ted, building scalable pipelines to ingest, process, and structure tender data across multiple procurement sources.
• built and improved search and retrieval systems using meilisearch, vector embeddings, reciprocal rank fusion (rrf), and llm-powered matching to improve contract discovery at scale.
• improved ai analysis pipelines for public tender documents, including document parsing, requirement extraction, dynamic translation, and structured llm-based insights, scaling processing to 1m+ documents.

ai engineer

audria | noida, india
jun 2025 - aug 2025

• designed and deployed an on-device ai agent for ios without external api dependencies, enabling orchestration across 20+ mcp servers and advanced tool-calling capabilities.
• optimised small language model (slm) inference on iphone hardware by leveraging apple's neural engine (ane), achieving a 30% speed improvement and reducing power consumption.
• built a custom automatic speech recognition (asr) pipeline tailored for ios, integrating voice activity detection (vad), speaker identification, diarization, and real-time speech-to-text (stt).

stack: swift, python, slms, stt, vad, diarization, mcp servers, ane, coredata, coreml.

ai/ml intern

imagined studios | remote
dec 2024 - apr 2025

• built leadgen, a centralized sales automation engine leveraging selenium and llm-based information extraction to convert unstructured web data into qualified leads, reducing lead qualification time by 70%.
• developed faq-ai, a no-code platform using flask that enabled clients to easily build rag-based ai chatbots with integrated speech-to-text (stt) and text-to-speech (tts) capabilities.

stack: python, flask, docker, cursor, llms, mongodb, chromadb, jwt refresh tokens, selenium, digitalocean, render, api testing and deployment.

ug researcher

the translational biology lab, indraprastha institute of information technology, delhi
dec 2023 - may 2025 | guide: dr. jaspreet kaur dhanjal

• conducted benchmarking of gpt-4o and domain-specific llms (e.g., biomistral, deepseek) on biomedical qa, using a corpus of 980 curated research papers and evaluation metrics including rouge-l, bleu, and meteor to ensure factual consistency and reliability.

ug researcher

etidm lab, indraprastha institute of information technology, delhi
aug 2023 - dec 2023 | guide: dr. kalpana shankwar

• developed a self-learning augmented reality (ar) system for precise pcb soldering by integrating computer vision techniques with unity and the vuforia engine, supporting real-time tracking and interactive tutorials.

publications

research

improving access to rare-disease knowledge: a retrieval-augmented question answering framework for wilson's disease
computers in biology and medicine, vol. 216, 111947 | 2026
tushar chandra, prateek paul, jaspreet kaur dhanjal
• introduces wilsonlitqa, a literature-grounded retrieval-augmented qa resource over the complete pubmed literature on wilson's disease. retrieval grounding substantially improves biological accuracy, and smaller ~7b-parameter models perform competitively when paired with a well-designed retrieval pipeline.
paper | pubmed | wilsonlitqa

abstract

objective: wilson's disease is a rare genetic disorder with a large but highly fragmented body of biomedical literature spanning more than a century of research. although thousands of articles are available through pubmed, this knowledge remains difficult to access in a consolidated, query-driven manner for researchers, clinicians, and students. existing large language models (llms) offer conversational access to information but often lack domain grounding, leading to hallucinations and unreliable responses in biomedical settings. in this work, we introduce wilsonlitqa, a literature-grounded retrieval-augmented question answering resource that enables the scientific community to query the complete pubmed literature on wilson's disease, capturing canonical biological, clinical, and pathophysiological knowledge related to the disease.

methods: our framework integrates hybrid information retrieval with generative language models, enabling users to query the wilson's disease literature as a unified knowledge resource rather than as isolated papers. we have systematically evaluated multiple open-source and proprietary llms under zero-shot and few-shot in-context learning settings, assessing their ability to deliver accurate, complete, and non-hallucinated responses when supported by retrieval.

results: our results demonstrate that retrieval grounding substantially improves biological accuracy and clinical relevance, and that smaller, deployable language models (on the order of 7b parameters) can perform competitively when paired with a well-designed retrieval pipeline.

conclusion: to support transparency, reproducibility, and community-driven development, the complete wilsonlitqa implementation is publicly available. together, this work positions retrieval-augmented question answering as a consolidated, continuously extensible knowledge interface for rare diseases, offering the scientific community a practical tool to navigate, interpret, and query the growing biomedical literature on wilson's disease.

education

academics

bachelors of technology in electronics and communication

indraprastha institute of information technology, delhi
aug, 2021 - july, 2025

my university offers the flexibility to choose courses from various disciplines, and i've primarily delved into interdisciplinary work with a strong emphasis on cs and ai fundamentals. few remarkable courses:

cs: • human-computer interaction (hci) • data structures and algorithms (dsa) • computer organisation (co) • advanced programming (ap) • machine learning (ml) • neuroscience of decision-making (ndm) • information retrieval (ir) • theories of deep learning (tdl) • computer vision (cv) • deep learning (dl)

maths: • linear algebra (m-i) • probability and statistics (m-ii) • multivariate calculus (m-iii) • differential equations (m-iv).

projects

work

loan.ly: ai lead generator for banks | demo
python, twilio, flask | vs code
• developed an automated cold-calling system that conducts intelligent loan and credit card application interviews, evaluates responses in real-time, and streamlines customer acquisition.

notegen: customized notes generator | demo
python, langchain, streamlit, conda, chromadb | vs code
• developed an ai-driven application allowing users to define topics and integrate custom knowledge bases for tailored notes generation. designed a flexible system architecture that supports markdown-formatted outputs, dynamic table of contents, and seamless integration of user-provided data pool.

prosearchai | demo
python, langchain, streamlit, conda, chromadb, gemini | vs code
• developed a semantic-based recommendation system that captures user intent by analyzing conversational context and product metadata, improving relevance beyond traditional keyword matches.

spambot: daily activity summarizer
python, streamlit, conda, gemini-pro | vs code
• spambot simplifies your life by summarizing daily activities from a transcript.txt file, providing both question-answering (qa) and summarization capabilities.

hindi translation - fre:ac (open source)
xml | vscode | guide: robert kausch
• added and tested hindi language support built from source.

...many more.

achievements

awards

team: spambots <3

hackathons

1st position in rajasthan digifest x tie global hackathon and a cash prize of rs. 75,000/-. (14, dec, 2025)
3rd position in nab fintech hackathon held at iiit-delhi. (16, feb, 2025)
1st position in 6th technovation hackathon held at sharda university under the theme of best use of ai in banking, and a cash prize of rs. 25,000/-. (26, jan, 2025)
1st position in neutron hackathon and a cash prize of rs. 50,000/-. (21, april, 2024)
won the best design award at red brick hacks held in ashoka university. (19, september, 2023)
won the best ai hack title at hackaithon held in shri ram group, jabalpur. (14, august, 2023)
1st runner up of green india hackathon, won a cash prize of rs. 10,000/- and an electric cycle worth rs. 40,000/-. (4, feb, 2023)
2nd runner up of india's biggest ev hackathon, decodev, won a cash prize of rs. 50,000/- and recognition in the parliamentary magazine, october. (11, september, 2022)

volunteering

volunteered at india's first-ever ea global conference in jaipur, and am a part of the ea community. (january, 2023)
core team member at young changemakers bootcamp'23 held at iit-delhi, a 5-day event in which we mentored 50 students from all over indian participated.

speaking & conferences

conference delegate at harvard's college project for asians & international relations (hpair'23) hong kong, china.
speaker at young changemakers bootcamp'23 held at iit-delhi, delivered a talk on "the art of storytelling".

other achievements

attended the 9th edition of ncbs's annual monsoon school, physics of life. a 10-day, in-person research symposium exploring biology through the lens of physics. exploring: non-linear dynamic system, active matter, flocking of birds, science communication (scicomm), development of inner ear in chicken embryo, marine particles & planktons, etc. (july, 2024)
semi-finalist in battle of bands of asia's biggest cultural event, mood indigo, held at iit bombay. (december, 2022)

let's talk

contact

not a founder but would love to ideate on interesting problem statements, both tech and non-tech, and create some impact together. always open to side projects, collaborations, and discussions.

drop a mail at: sde.tusharchandra@gmail.com

keep hustling | good vibes only :))