Experience

• Je tra­vaille actuelle­ment sur l’in­fra­struc­ture pour les flux de tra­vail agen­tiques et l’­analyse des re­tours. Mon équipe a tra­vaillé sur les chat­bots d’IA pour les mes­sageries pro­fes­sion­nelles, ce qui a con­duit à la créa­tion d’une ar­chi­tec­ture stan­dard­isée per­me­t­tant d’exé­cuter ces flux de tra­vail. Elle est ca­pa­ble de gérer de grandes quan­tités d’ap­pels agen­tiels. Le cadre sous-ja­cent est Pydantic AI.

• My team im­ple­mented and sup­ported the Business Messaging prod­ucts at Partoo.

  www.par­too.co/​en/

• WeLearn - A tool for col­lec­tive learn­ing

  • Responsible for ar­chi­tect­ing and de­vel­op­ing APIs for a Machine Learning model that clas­si­fied web­pages into Wikipedia Pages (concepts).
  • Created vi­su­al­i­sa­tion (a map of con­cepts) that ef­fi­ciently dis­plays a users concepts with spa­tial cues and the abil­ity to find other sim­i­lar con­cepts.
  • Developed browser ex­ten­sions to in­gest web­pages.

  www.learn­ing­plan­etinsti­tute.org/​en/

• Hydrodynamics of Super-Paramagnetic Micro-Beads

  My work here was to cap­ture the mo­tion of the mi­cro beads un­der­go­ing at­trac­tion due to an ex­ter­nal mag­netic field. Based on this it is pos­si­ble to quan­tify hy­dro­dy­namic drag forces on the beads.

  Using open source tools trackpy and scikit-image, I de­vel­oped an im­age pro­cess­ing pipeline to iso­late, track, and cal­cu­late mo­tion dy­nam­ics from the frames. We reached an av­er­age frame rate of ~800 fps.

  www.pmmh.es­pci.fr

• The Mechanical Sense of Paramecium

  As a re­search in­tern, I de­vel­oped high-fre­quency, real-time im­age analy­sis and hard­ware con­trol pipeline to track long tra­jec­to­ries of Paramecium. I wrote a dri­ver for the mi­cro­scope stage, and a high-speed im­age aqui­si­tion and pro­cess­ing (~50fps) tool. Using these soft­ware we were able to fol­low the tra­jec­tory of mov­ing Paramecium.

  www.la­bos.upmc.fr/​ljp

• Study of Diversity in Phyllotactic Patterns

  Phyllotaxis is the arrange­ment of lat­eral or­gans in plants. In this pro­ject we at­tempt to ex­plain the ori­gin of emer­gent pat­terns, tran­si­tions, and ir­reg­u­lar­i­ties in such arrange­ments. Using cylin­dri­cal pro­jec­tion un­rolling and un­warp­ing, I de­vel­oped an im­age analy­sis pipeline al­low­ing bet­ter study of com­pli­cated plant or­gans.

  www.msc.univ-paris-diderot.fr

• Personalised Medical Events, Conferences, and Courses Recommendation for Doctors at NHS, UK

  • Developed the al­go­rithm for clas­si­fi­ca­tion of events into med­ical spe­cial­i­ties.
  • Integrated the clas­si­fi­ca­tion sys­tem into the web app.
  • Implemented an event rec­om­men­da­tion en­gine lever­ag­ing Elasticsearch and stored fea­tures.
  • Wrote a frame­work for event data ag­gre­ga­tion from mul­ti­ple NHS web­sites.

  ago­ra­health.org.uk

• Computer Vision and Information Extraction

  Developed and shipped a util­ity for dig­i­tiz­ing wave­forms and ex­tract­ing meta­data from ECG re­ports. It is be­ing used at Brigham and Wom­en’s Hospital, Boston for pro­cess­ing more than a mil­lion pa­tient records.

  Also worked on ex­tract­ing nor­mal­ized tab­u­lar data from scanned med­ical records. Key chal­lenges here in­volved han­dling skewed, de­sat­u­rated and oth­er­wise low qual­ity im­ages.

  oxyent.com

• Survaider Web App and Survey Builder

  Developed a web app back­end and REST API in Python.

  Designed and de­vel­oped a web based sur­vey builder tool which was later in­te­grated into the core ap­pli­ca­tion.

  sur­vaider.com

• Human Activity Recognition from Acceleration data Under Naturalistic Conditions

  The goal of this pro­ject was to de­velop a sys­tem for recog­ni­tion of low level hu­man ac­tiv­i­ties, such as Walking, Running, Sitting, etc., us­ing a sin­gle ac­celerom­e­ter.

  Using su­per­vised learn­ing method, we trained a Random Forest Classifier us­ing fea­ture vec­tors ob­tained via our model. We were able to pro­duce an ac­cu­racy of 95% through our model.

• Aspiring Researcher Challenge

  As a re­search as­so­ci­ate, I was part of the team which was re­spon­si­ble for de­vel­op­ing a large, on­line ex­per­i­ment to study the wis­dom of the crowd where I for­mu­lated the al­go­rithm for scrap­ing and for­mat­ting raw data off mul­ti­ple crowd­fund­ing plat­forms.

• Modelling and Network Analysis of Intronic miRNA Mediated Gene Expression Regulation

  In this pro­ject, we de­vel­oped a math­e­mat­i­cal model for gene ex­pres­sion reg­u­la­tion via in­tronic mi­cro RNAs, based on em­per­i­cal data and ex­per­i­men­tal ob­ser­va­tions at IGIB.

  As a part of study, we’ve also gen­er­ated a miRNA <=> mRNA tar­get net­work us­ing var­i­ous avail­able data­bases.

• Developed an open-source and mod­u­lar ERP sys­tem and API sub-sys­tem.

• Built a clas­si­fi­ca­tion sys­tem for geospa­cial fea­tures be­tween texts orig­i­nated in dif­fer­ent con­ti­nents. Other re­spon­si­bil­i­ties in­cluded de­sign and de­vel­op­ment of sev­eral web por­tals, graphic de­sign­ing, me­dia and pub­lish­ing tasks.