Dr. Gunjan Gupta & Amit Singh, co-founders of Lightsphere AI Inc, a Bellevue, WA based AI company, gave an introductory talk at the 2018 Parent University event at Tyee Middle School on the future of careers in the age of Advanced AI and how young people entering into the process of selecting their tracks for a career need to think in order to keep up with the rapidly changing job market.

This was a talk by invitation from the Tyee school and PTSA, and was part of a series of seminars that included other speakers.  The presentation focused on issues and advice for college admissions for students in eighth grade all the way up to college level. The session was especially designed to showcase and explain how an “early start in self-exploration for possible career options” can have a huge effect on successfully finding a good matching program for a productive and satisfying long-term career. Important information was shared about how rapid changes in the job market due to automation and shifting demographics are requiring a change in the way students need to think, explore, apply and pursue post-secondary education for success.

The workshop was attended by around 52 students and parents and was very well received, both by parents and the PTSA; thanks a lot to Tyee Assistant Principal LeAnn Tuupo, Tyee PTSA President Mei Chen, BSF Ambassador Rob Fatlan, and the rest of the organizers and PTSA board members for their support.

We received great reviews from the talk and a lot of interest for a follow-up session for the first time, we shared with public, and especially students who are about to enter high-school, a very important discovery from our research on AI to help students that is likely to change the way we think about education and career.

The discovery we presented is closely tied to our suberin product; over the coming weeks we will be conducting summer camps in Seattle and Chicago area, where we would go even deeper in sharing some of the really cool tech our team is building.

98.7% accuracy now in matching emails to authors…

Lot’s of calculations below to figure out how many authors we can actually associate with emails correctly with the correct author name since they appear in crazy variations…

See number in bold below.  When we started today, the accuracy was 79%…

Next step; author identity timeline construction..
-Gunjan Gupta


And paper count up to 11k almost…

; 570/720
; 115/(115+25)
; {‘graph_agree_identity’: 554, ’email_zip_corrections’: 362, ‘identity_missing_graph’: 4, ‘graph_email_author’: 706, ‘known_email_author’: 11543, ‘total_email_author’: 11547, ‘pop_email_author’: 724, ‘graph_disagree_identity’: 148}^C

; 29/(8+29)
; 609/706
; (706-609)*0.78
; ((706-609)*0.78+609)
; ((706-609)*0.78+609)/706
; ((706-613)*(31/37)+613)/706
; exit
gunjan@sandgun11:~/pr_data/uber/Computer_Science/Machine_Learning/pdf_info/extracted$ calc
C-style arbitrary precision calculator (version
Calc is open software. For license details type: help copyright
[Type “exit” to exit, or “help” for help.]

; ((706-615)*(45/50)+615)/706
; exit

{‘graph_agree_identity’: 615, ’email_zip_corrections’: 1074, ‘identity_missing_graph’: 4, ‘graph_email_author’: 706, ‘known_email_author’: 11543, ‘total_email_author’: 11547, ‘pop_email_author’: 724, ‘graph_disagree_identity’: 87}
[Computer Science.Machine Learning]: [7. TR CY paper entities]: COUNT: total:35, unshipped: 0, processed: 35, processed_file_count: 35, error: 0
Entity Extraction Metrics for all: {
“author_less_than_email_count”: 147,
“email_count”: 19155,
“paper_count”: 10705,
“{CONFERENCE}_global”: 3,
“{DEPARTMENT}_global”: 28,
“{UNIVERSITY}_global”: 24

(812) 890-7564

And these were extracted using our AI that reads author papers and other websites, not human labor. See stats below. Getting close to the first major boost to universityMatch.AI beta released earlier.


[Computer Science.Machine Learning]: [7. EX CY paper entities]: COUNT: total:29, unshipped: 0, processed: 29, processed_file_count: 29, error: 0
Entity Extraction Metrics for all: {
“email_count”: 16067,
“extract_count”: 9541,
“fail_count”: 0,
“info_paper_count”: 9586,
“missing_pdf_count”: 45,
“pdf_paper_count”: 9541,
“{CONFERENCE}_global”: 3,
“{DEPARTMENT}_global”: 28,
“{UNIVERSITY}_global”: 24

universitymatch.AI moving quickly towards production, preview and beta coming soon!

We have made great strides towards production; the sneak-peak is coming out this Friday where you will be able to actually login (for the first few select sign-ups) and walk through Computer Science – Machine Learning rankings, colleges, and professors. We still expect to go full public beta on June 15 for every early-bird who pre-orders when all of Computer Science and the new exploration interface will be available. We plan on going to full production on July 4, at which point the price will revert to full.

Note that even by late May when we plan on doing another campaign and the beta might already be available for users, the price is no longer going to be as low as it is now, and will revert substantially closer to the $100 per year target price for us for 2018.

(812) 695-0117

Just a quick heads up that we will be launching an interesting Instagram campaign for Deepa Dance on Monday. It’s already up here right now, so if anyone wants to leverage it, you can do so right now by clicking and buying that Perk from our 870-454-2964. Great way to get a poster signed by an awesome Bollywood star and also a 1 year membership to universitymatch.AI.

BTW, check out this very cool new video that highlights in a short one minute why we need universitymatch.AI.