From Bangalore biofuel labs to Kenyan savannahs to Cambridge pharma - follow the path.
The Journey
Research Positions
Click any milestone to expand.
2023
Biotech R&DR&D Research InternSea6 Energy · Bangalore, India
Summer 2023
50+ PCRs
Bangalore, India · Seaweed biofuel R&D
Studied Seaweed Carrageenan bioconversion into biopolymers, biostimulants, and platform chemicals alongside a team of PhDs and postdocs
Conducted 50+ PCRs for bacterial infection models, HPLC molecular analysis, and evaluated Carrageenan with 5 additives across 9 plant species
NeuroscienceComputational Neuroscience InternDevilbiss Lab · Rowan-Virtua SOM
Jun 2023 - Aug 2024
8 subjects
3-ch EEG · TBI · LC-NE research
Research on locus coeruleus-norepinephrine (LC-NE) regulation of neural activity and cognition using 3-channel EEG on 8 animal subjects (4 rabbits, 4 rats) with and without TBI
Engineered EEG artifact-removal pipelines: band-pass filtering, ICA for ocular/EMG artifacts, and threshold-based epoch rejection prior to spectral analysis
Measured neural dynamics with norepinephrine, dopamine, and Ritalin interventions; analyzed with MATLAB and SpikeOne; directed toward SOM/GSBS publication
2024
ClinicalNeurology Medical Assistant & ShadowingComprehensive Neurology LLC
Summer 2024
50+ patients/day
MRI · EEG · EMG · 8 weeks full-time
8-week program, 40 hrs/week; aided 50+ patients/day and managed WC, MVA, NJ Disability, and surgical precertifications using Practice Fusion EHR
Shadowed Drs. Pasupuleti & Zhang - observed cases spanning Parkinson's, Alzheimer's, Epilepsy, MS, and motor neuron disease; interpreted MRI, EEG, and EMG results
NeuroAINeuroAI Research InternHasson Lab · Princeton University
Oct 2024 - Jun 2026
1.5M+ hrs
$50M NIH · 15 families · infant cognition
Analyzing 1.5M+ hours of infant cognitive data from 15 families in the 1,000 Days (1kD) Project - a $50M NIH-funded longitudinal study of early brain development
Applying Python-based deep learning pipelines and ACLEW tier-dependent annotation frameworks to quantify executive function (EF) and self-regulation markers
Filtering confounds and standardizing annotation tiers to build scalable cross-family predictive models for infant neurodevelopment
2025
Field ResearchResearch Assistant - Grevy's Zebra ProjectMpala Research Centre · Laikipia, Kenya
Jan 2025 - Present
20+ hrs 4K
Laikipia Plateau · FAA + Kenya certified
Examining flee dynamics of Plains and Grevy's Zebras in response to predation pressure using drone footage captured from ~100m altitude over the Laikipia Plateau, Kenya
FAA & Kenya-certified drone pilot; captured 20+ hours of 4K footage during a 3-week field collection sprint
Applying DeepLabCut and SLEAP to map individual and group-level movement patterns; collaborating with Dr. Craig Marshall and Dr. Isla Duporge on a manuscript for The Naturalist
Clinical ResearchResearch Fellow - Undergraduate Clinical Scholars ProgramUPenn Perelman School of Medicine
Summer 2025
500+ patients
43% outcome improvement · FMT study
Analyzed 500+ patient microbiome datasets in R, Stata, and Excel to evaluate FMT route of administration for severe recurrent Clostridioides difficile infection
Identified 43% improvement in patient outcomes with lower GI (enema) FMT vs. upper GI routes; presented to 100+ senior physicians at UCSP symposium
Authoring 5 conference abstracts and 2 peer-reviewed manuscripts; completed NIH/NIDDK-backed epidemiology training
AI / MLAI/ML DeveloperPEMM Department · Princeton University
May - Aug 2025
4,500 paintings
DELLA supercluster · BLIP-2 · CLIP · CoCa
Built computer vision pipelines on Princeton's DELLA supercluster to analyze 4,500 Ethiopian paintings and 200,000+ metadata attributes for large-scale art-historical analysis
Adapted and fine-tuned BLIP-2, CLIP, and CoCa vision-language models to build a scalable IC|TC classification framework for automated stylistic attribution (First vs. Second Gondar styles)
Data AnalysisData Analyst - Project XenonAfterQuery Experts
May - Aug 2025
100+ datasets
~80% model accuracy · 1,000+ annotations
Analyzed 100+ clinical and economic datasets in Excel, R, and SQL for downstream decision modeling
Built labeled training frameworks that improved AI model task performance to ~80% accuracy via statistical modeling, feature engineering, and 1,000+ quality-controlled annotations
VolunteerCare Ambassador - Neurology DepartmentPrinceton Medical Center · Penn Medicine
Sep 2025 - Present
150+ hrs
Live hospital neurology · Penn Medicine
150+ volunteer hours supporting nursing workflows, patient interaction, and interdisciplinary clinical teams in the Neurology Department at Princeton Medical Center
2026
Public HealthLead Data Analyst - NJ Mobile Harm Reduction StudyPrinceton Global Health Program
Feb 2026 - Present
574 responses
14 NJ sites · 28 policy visualizations
Lead analyst on a 574-response client survey across 14 New Jersey mobile harm reduction sites - PI: Prof. Heather Howard (SPIA); Lead: Dr. Ben Scher
Built the full R/Python pipeline: substance-use profiling, naloxone access cascades, police-interaction modeling, housing-instability risk - surfacing significant racial disparities in naloxone access (OR = 1.74, p < 0.05)
Produced 28 policy-facing visualizations informing NJ statewide harm-reduction service-delivery recommendations
NeuroscienceJunior Paper & Senior Thesis ResearcherPrinceton Neuroscience Institute · Brody Lab (Carlos Brody, HHMI)
July 2026 - Present
Neuropixels
mPFC + hippocampus · learning · Princeton
Modeling how prefrontal and hippocampal populations reorganize as mice learn a virtual-reality decision task, from chronic Neuropixels recordings in 3 implanted animals tracked across every stage of learning
Built the analysis pipeline from single-cell GLMs to population flow-field models (neural SDE + GRU) to capture how task solutions form
Adding closed-loop optogenetic hippocampal inhibition to test its causal role in shaping prefrontal dynamics, mirroring adaptive DBS
Basis of the junior paper and senior thesis, mentored by Jesse Kaminsky
Built the analysis and review platform behind the lab's full Percept dataset, 202 deep-brain recordings across 87 Parkinson's and dystonia patients, now the tool the team uses to catalogue and annotate every session for adaptive DBS biomarker discovery
Showed with patient-nested mixed models that beta power falls as stimulation amplitude rises (-0.90 dB/mA, p < 1e-12), confirming across all 87 patients that the control signal adaptive DBS depends on holds beyond single-patient studies
Ran a scan across all 202 recordings for biomarkers invisible to visual review and surfaced the aperiodic 1/f slope, which tracks stimulation second only to canonical beta-burst rate (|r| = 0.58 overall, 0.79 in GPi)
Invited to present the findings to the Starr Lab in September 2026
Built an end-to-end gene evidence platform in 8 weeks, unifying 20 population-scale genetic databases through Novartis's internal warehouse to surface and rank drug targets
Now used daily by the DAx team for gene safety screens and more broadly for target discovery and portfolio triage, replacing a multi-day manual evidence search with an on-demand query
Extended it into an agentic LLM system with a hard cite-or-abstain guarantee, where every numeric claim must appear verbatim in its cited source row or the system declines, proven by 1,027 adversarial checks with zero fabrications
Found that default query row caps silently dropped 60-94% of the genetic evidence on high-degree genes, and shipped a completeness-preserving fix (see the architecture)
Scholarly Work
Publications in Progress
Seven manuscripts and abstracts - clinical medicine, neuroscience, ecology, genomics, computational art history.
01
Fecal Microbiota Transplantation Route of Administration and Clinical Outcomes in Recurrent Clostridioides difficile Infection
Donthula A. et al.UPenn Perelman · UCSP 2025-2026Under Review
Optogenetic Sweet Neuron Activation Supports Spatial Place Conditioning in Drosophila melanogaster, with Preliminary Evidence for Goal-Directed Outcome Sensitivity
Donthula A.NEU 350 · Princeton University · May 2026In Preparation