You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
$1,020 left to reach the funding goal.
i’m sushanth, a high-school student from india, and over the past several months i’ve been independently working on research about ai agents and the enormous amount of execution data they generate.
the result was a paper called “the trace corpus is mostly redundant: measuring, compressing, and exploiting agent-execution data at scale.”
a few days ago, it was accepted as a regular paper at ieee bigdata 2026.
the conference received 1,046 submissions and accepted 203 regular papers, an acceptance rate of 19.4%. getting that email was probably one of the biggest moments i’ve had as a student researcher. i don’t work out of a university lab, i don’t have a professor’s grant paying for my experiments, and i don’t have a department conference budget behind me. most of this work was done independently because i genuinely wanted to understand the problem and see whether the idea held up.
the research asks a fairly simple question: when ai agents perform long tasks, how much of everything they record is actually useful?
agent runs can produce huge traces containing actions, intermediate states, tool calls, observations, retries, and other execution data. we studied how redundant these traces are, how much they can be compressed, and whether the compressed data still preserves what matters for downstream analysis and learning.
the research is finished. it passed peer review. the paper has been accepted.
the problem is much less interesting, i can’t afford the publication fee.
ieee bigdata requires at least one author of every accepted paper to purchase a full author registration. because i am not currently an ieee member, mine is $1,020.
i’m going to present remotely, so i’m not asking for flights, a hotel, food, or travel expenses. this entire request exists for one reason: to pay the registration fee required for an already accepted paper to enter the conference proceedings.
if this gets funded, $1,020 turns completed research and an acceptance email into an actual published ieee paper.
the immediate goal is very concrete: get the accepted paper through the final publication process.
i’ll incorporate the reviewers’ feedback, prepare and submit the camera-ready manuscript, complete the required presentation video, and register the paper before the conference deadline.
beyond simply publishing it, i want the work to be useful to people building and studying ai agents. agent systems are producing larger and larger execution datasets, and if much of that information is redundant, understanding what can safely be removed could make these datasets substantially cheaper to store, process, and learn from.
the research itself is already completed and accepted. funding is the remaining bottleneck.
there is a very clear success condition.
the $1,020 author registration is paid before the deadline, the camera-ready paper and presentation are submitted, and the accepted paper proceeds into the ieee bigdata 2026 proceedings.
there isn’t really an ambiguous outcome here. either the registration barrier is removed and the accepted work gets published, or it isn’t.
100% of the requested $1,020 will go toward the ieee bigdata 2026 non-member author registration fee.
registration: $1,020
travel, hotel and other expenses are all $0
i will be presenting remotely. if i find a legitimate way to reduce the registration price before payment, such as qualifying for the ieee member rate or receiving conference support, i will update the proposal rather than treating the difference as extra funding.
i’m sushanth dasari, a high-school student researcher from india.
i spend a large part of my time independently working on problems in ai, machine reasoning, evaluation, and scientific machine learning. i usually start with a question i don’t know how to answer, build the experiments needed to test it, and keep iterating until either the idea fails or there is something worth writing down.
this project followed that process. i developed the research, built and ran the experimental work, analysed the results, wrote the manuscript, submitted it for external peer review, and it was ultimately accepted as a regular paper at ieee bigdata 2026.
this is not funding for an idea i hope to start someday. the work already exists and has already passed the conference’s review process. i’m asking for help with the final step required to make it public.
the main way this fails is simple: i don’t raise enough to pay the mandatory author registration before the deadline.
the paper itself has already been completed and accepted, so the technical risk is largely behind me.
the unfortunate outcome would be having research successfully pass peer review but being unable to complete publication purely because, as a high-school student without institutional funding, i cannot personally cover a $1,020 conference registration fee.
that is exactly what i’m trying to avoid.
for this publication cost $0.
i currently do not have a university research budget, advisor grant, departmental conference fund, or other institutional funding available to cover this registration.
i am also asking the conference whether any financial assistance is possible, and if another source covers part or all of the registration before this grant is used, i will update the proposal accordingly.
for the form itself, i'd set minimum funding = $1,020, maximum funding = $1,020, and the decision deadline around november 10, because the fields currently shown in your manifund form are $500/$500 with a november 21 deadline, which would be too late for the ieee deadline.