You can create a custom user-defined aggregation function (UDAFs), currently in preview:

CREATE OR REPLACE AGGREGATE FUNCTION `project.dataset.latest_non_null`(
  value STRING,
  some_timestamp INTEGER)
RETURNS STRING
AS (
  ANY_VALUE(value HAVING MAX if(value is null, null, some_timestamp))
);

SELECT
  a,
  project.dataset.latest_non_null(b, some_timestamp) AS latest_b,
  project.dataset.latest_non_null(c, some_timestamp) AS latest_c,
FROM
  foo
GROUP BY
  a

Output: