Hackathon - Instructions document (1)


Arav1201

Uploaded on Feb 26, 2026

Comments

                     

Hackathon - Instructions document (1)

CloudCosmosHackathon–On-Campus Recruitment Round ThishackathonisthefirststepoftheCloudCosmosrecruitmentprocess.Youwillbuild end-to-endAI- drivensolutions,submittheworkviaGitHubrepositories,anddemonstrateyourproblem- solving and implementation skills. HackathonProblemStatement—“Clone&Elevate” Brief Pickawell-knownproductideaandbuildafocusedcloneorre-imaginationofit.Youarenot limitedtoYCombinatorstartups,chooseanystrongreference(e.g.,acustomLLM,anAIcode editorlikeCursor,anAInote-taker,aRAG-basedknowledgeassistant,etc.)oryourownidea from scratch. Yourgoal:aworkingprototypewithsolidengineering,thoughtfulscope,andclearimpa ct. ProblemStatement&IdeaConfirmation StudentAction Studentsmust: ● ChoosetheirideaandsubmititviaGoogleForm. NOTE:Formwillbeopenfor2hoursonly. FormTitle:CampusHiringDrive–CloudCosmosTechnologiesLLP:IdeaSubmission FormFields: ● Name(required) ● Email(required) ● RollNumber(required) ● RegistrationNumber(required) ● GitHubProfileURL(required) ● GitHubRepository(publicrepository hereyou’llpushthecode)(required) ● Problem Statement( hat you ill be building, share your idea) (required) ● PhoneNumber(required) FormURL:https://forms.gle/dSYKaoid5NaePHMNA CloudCosmosAction ● CloudCosmoswillreviewandconfirmthefeasibilityofeachidea. ● Ifitdoesn’tmeettheguidelinesorrequiresanychanges,amemberofourteamwill reachouttoyouwithinthenext30minutes.Ifyoudonotreceiveanyfollow-upemail,it means your problem statement has been accepted, and you may go ahead and startbuilding. Build&Submission BuildWindow 🕗Oct24(8:00AMIST)→Oct26(8:00PMIST)—acontinuousbuildwindow. You are expected to complete and submit the projects within this timeframe. Deliverables 1. Sourcecoderepositorylink(GitHub). ○ Repositoriesmustbepublic forCloudCosmosreviewers. 2. Demovideo(≤3minutes)uploadedtoYouTube. ○ AddtheYouTubevideolinktoyourrepositoryREADMEandthesubmissionform. 3. (Optional)DeployedbuildonRender,Vercel,Netlify,orsimilarplatformforlivedemo access. SubmissionMethod SubmityourfinalprojectviatheFinalBuildSubmissionGoogleForm: FormTitle:CampusHiringDrive–CloudCosmosTechnologiesLLP:FinalBuildSubmission FormFields: ● Name(required) ● Email(required) ● RollNumber(required) ● RegistrationNumber(required) ● GitHub RepositoryURL( hereyourcode ispushed)(required) ● YouTubeURL(recordeddemouploadedonYouTube)(required) ● DeployedapplicationURL(Optional) FormURL:https://forms.gle/E1vaLHZ1fCU7cumC8 QuerySupportWindow(Important) TheCloudCosmosSupportTeamwillbeavailabletoanswerqueriesfor5hoursimmedi ately after the problem statement is shared. ⏰Noqueries illbeentertainedbeyondthis5-hour indo . Pleaseensurequestionsareclear,concise,andrelevant.Unnecessaryorrepeatedquerieswill not be addressed. InstructionsonHowtoRaiseaQuery We’vedividedparticipantsequallyamongsixPOCs. Whenyouraiseanyqueriesduringthehackathon,youhavetocctheirrespectivePOCwhi leemailing the centralized support address:punjabi-univ-placements- [email protected] AsheetlistingeachgroupandtheirassignedPOCisattachedwithemail. Example: If,forexample,SanamhasaqueryandherassignedHackathonSupport(PointofContactis Baval, she should send her email to: 🕗**punjabi-univ-placements- [email protected]**CC:Baval’s(email as mentioned in the attached sheet) ThisensuresthequeryisvisiblebothtothecentralsupportandtheassignedPOCforfasterresolutio n. Evaluation&Results EvaluationCriteria(Total:100pts) Category Weig ht Innovation&ProblemUnderstandin 25 g TechnicalDepth&CodeQuality 25 Feasibility&Completeness 20 Impact&Scalability 15 Presentation&Communication 15 ResultsAnnouncement - Resultswillbeannouncedwithinaweekpostsubmissions. - Resultswillbeannouncedviaemailtocandidatesandtheplacementcell. - SelectedcandidateswillproceedtoRound2oftheCloudCosmoshiringprocess. ParticipationGuidelines ● TeamSize:Individualparticipationonly. ● Eligibility:Registeredstudentsofthe campusdrive. ● AllowedTech:Anyopen-sourcetools,libraries,orframeworks. ● OriginalWork:Mustbeyourown.AI-assistedcodingispermittedwithcitationoftools used (e.g., ChatGPT, GitHub Copilot, etc). ● CodeofConduct:Maintainprofessionalismthroughoutthehackathon. WhattoBuild(ScopeSuggestions) ● MVPfirst:Focusononeortwostrongfeaturesthatdemonstratevalue(e.g.,“Ask- DocsRAGwithcitations,”“Inlinecodeeditsviaagenttools,”“Fine- tunedmodelfordomainX”, etc). ● Depthoverbreadth:Onepolishedfeature>fiveincompleteones. ● Elevatethereference:Don’tjustcopy:ImproveUX,reliability,orperformance,etc . MinimumTechnicalRequirements ● PublicGitrepositoryfromthestart(READMEmustincludesetupandrunsteps). ● Runnable,accessiblebuild:judgesshouldbeabletorunorviewiteasily. ● MandatoryYouTubedemo(2–3mins)showingkeyfeaturesandflow. ● Ifusingexternalmodelsordatasets,citesourcesandmentionlimitations. ● AnyLLM(OpenAI,Claude,Gemini,Llama,Mistral,etc.)maybeused:stateyourchoicein README. RepositoryNamingFormat /- Example: ai-code-editor/rahul-sharma-2201 OptionalTechHints Backend/API:Python+FastAPI RAG/VectorDB:FAISS,OpenSearch,Weaviate,Milvus Agents/Orchestration:OpenAISDK,AutoGen,function-callingtools ML:PyTorch,TensorFlow,Scikit-learn Cloud/Infra:Docker,basicCI,simpleclouddeployment SampleREADMEformat Youarenotstrictlyrequiredtousethisexacttemplate.However,yourproject’sREADMEshould follow asimilar structure and level of detail. ##ProjectOverview --- ##ProblemStatement Brieflydescribetheproblemyourprojectaddressesandthemotivation behind it ( Example: > "Students face difficulty organizing research notes across multiple formats.WebuiltanAI-powereddocumentsummarizerthathelpsmanageand query large sets of academic documents using RAG." --- ##SolutionSummary Provideashortexplanationofyourideaandhowitworks. > Example: > "WebuiltalightweightAIwebappthatletsusersuploadfilesandask contextual questions. > ItusesFastAPI+FAISSforretrievalandOpenAI'sAPIforanswergeneration." --- ##TechStack Listalltechnologies,frameworks,APIs,andtoolsyouused. - **Backend:**Python,FastAPI - **Frontend:**React/Streamlit/FlaskUI/HTML-CSS(chooseasapplicable) - **Database/VectorDB:**PostgreSQL/FAISS/Weaviate/Milvus - **LLM/AIModels:**OpenAIGPT-4/Claude/Gemini/Llama/Mistral - **Cloud/Hosting:**Render/Vercel/HuggingFaceSpaces/Netlify/ Glitch - **VersionControl:**Git+GitHub --- ##ProjectStructure ```root/ ├──app/ #corebackendormainsourcecode ├──s t a t i c / orfrontend/ #frontend/staticfiles ├──requirements.txt #alldependencies ├──environment.yml #condaenvironmentfile(optional) ├──README.md #thisfile └──d e m o . m p 4 (optional) #localvideoreferenceifany ``` --- ##SetupInstructions(withConda) Followtheseexactstepstorunyourprojectlocally. ```bash #1.Clonetherepository gitclonehttps://github.com/.git cd #2.Createandactivateacondaenvironment conda create -n cloudcosmos python=3.10 -y conda activate cloudcosmos #3.Installdependencies pipinstall-rrequirements.txt #4.Runtheapp python app.py # OR (if using FastAPI) uvicornapp:app--reload ``` Ifyou'reusingaseparatefrontend,mentionsetupandstartcommands below: ```bash #ForReact(example) cd frontend npminstall npm start ``` --- ##Deployment **LiveDemoURL:** `https://your-deployed-app-link.com` > Ifnotdeployed,mentionthattheappcanberunlocallyusingthesetup steps above. --- ##DemoVideo(Mandatory) **YouTubeLink:** `https://youtu.be/your-demo-link` > Thedemovideo(2-3minutes)mustclearlyshow: > -Howtheappworks > -Corefunctionalityandflow > -Anyuniqueorinnovativefeatures --- ##Features Listthemainhighlightsofyourproject(3-5points). - End-to-endworkingwebapp(accessibleviabrowser) - ImplementsAI/LLM-basedlogic - CleanUIandstableperformance - Real-worldapplicability - Clear,maintainablecodestructure --- ##🕗TechnicalArchitecture Explainbrieflyhowyoursystemworks. AsimpletextdescriptionorASCIIdiagramissufficient. > Example: > -Frontendsendsquery→B a c k e n d API(FastAPI) > -BackendfetchesvectorresultsfromFAISS > -Sendscontext+querytoLLM→R e t u r n s summarizedresponse > -Responsedisplayedinwebinterface --- ##🕗T e st i n g (Optional) Ifyouaddedtestsorvalidationsteps,includedetailshere. ```bashpytesttest s/ ``` --- ##🕗References Listanyopen-sourceresources,datasets,ormodelsyouused. - [OpenAIAPI](https://platform.openai.com/) - [LangChainDocumentation](https://python.langchain.com) - [HuggingFaceDatasets](https://huggingface.co/datasets) - [FAISSLibrary](https://github.com/facebookresearch/faiss) --- ##⚖License(Optional) Youcanincludealicenseifyourprojectusesone.Example: `ThisprojectislicensedundertheMITLicense.` --- ##🕗 Acknowledgements Mentionanylibraries,ortoolsthathelpedyoubuildthisfaster. > Youarenotstrictlyrequiredtousethisexacttemplate--however,your project'sREADMEshouldfollowa**similarstructureandlevelofdetail.** FinalNote Goodluck—shipsomethingsmall,sharp,andimpressive! Focus onclarity, execution, and reliability. Makeityourbestdemonstrationofreal-worldengineeringskill.